1
00:00:00,240 --> 00:00:02,080
Very excited for today's 
episode. 

2
00:00:02,280 --> 00:00:04,480
I'm excited for every episode, 
but we should have a great 1 

3
00:00:04,480 --> 00:00:06,040
today. 
It's a full house. 

4
00:00:06,040 --> 00:00:08,760
It's me, Tom Doton, Eric 
newcomer Mad, Madeline 

5
00:00:08,760 --> 00:00:13,240
Renvarger, and our special guest
Ed Zittron of the Better Offline

6
00:00:13,240 --> 00:00:16,320
podcast of the Better Off. 
Ed, is that the name of your? 

7
00:00:16,360 --> 00:00:17,840
So where's your Ed at? 
Where's your Ed? 

8
00:00:17,840 --> 00:00:18,960
I'm sorry. 
Where's your Ed? 

9
00:00:19,720 --> 00:00:25,080
Ed was like an NBC show. 
So is Basement Jack's reference.

10
00:00:25,560 --> 00:00:28,680
Yeah, OK. 
And and also, I mean in in 

11
00:00:28,680 --> 00:00:30,760
fairness to to all of this, you 
also is it you're you're still 

12
00:00:30,760 --> 00:00:33,360
working at least somewhat on the
side doing some, some comms for 

13
00:00:33,400 --> 00:00:35,400
I'm I'm run. 
PPR from my day job. 

14
00:00:35,400 --> 00:00:38,160
I just keep it firewall and I 
don't cover anything close to 

15
00:00:38,160 --> 00:00:39,120
what I do. 
Got it. 

16
00:00:39,400 --> 00:00:41,720
OK, OK, great. 
But you've also, you know, 

17
00:00:41,720 --> 00:00:43,360
you've been on a tear. 
You've you've been around tech 

18
00:00:43,360 --> 00:00:45,240
for a long time. 
As long as I've been around it, 

19
00:00:45,480 --> 00:00:47,760
yeah. 
And you know, at various times I

20
00:00:47,760 --> 00:00:51,120
think really step up to call 
bullshit on what you find to be 

21
00:00:51,280 --> 00:00:54,640
disturbing and and kind of built
into the core of the tech 

22
00:00:54,640 --> 00:00:57,360
industry and the hype machine. 
But you've really been on a tear

23
00:00:57,520 --> 00:00:59,920
in the last, I mean probably a 
couple years, but definitely the

24
00:00:59,920 --> 00:01:03,400
last few weeks over generative 
AI and you have published the 

25
00:01:03,400 --> 00:01:05,519
haters guide to the generative 
AI bubble. 

26
00:01:05,560 --> 00:01:08,400
You have written about the 
insitification of AI. 

27
00:01:08,960 --> 00:01:10,880
Yeah, we're we're going to have 
a lot to dig into here, but I 

28
00:01:10,880 --> 00:01:13,000
want to give you the space to 
kind of lay out your argument. 

29
00:01:13,040 --> 00:01:17,040
And before I do that or and 
leading to that, I should say I 

30
00:01:17,040 --> 00:01:19,440
want to read something that 
maybe outlines kind of very 

31
00:01:19,440 --> 00:01:23,160
broadly from your haters guide. 
And this I think just kind of 

32
00:01:23,160 --> 00:01:26,240
gives a summation of why you're 
so mad about the state that 

33
00:01:26,240 --> 00:01:28,760
things are in within the the 
tech world. 

34
00:01:28,760 --> 00:01:31,640
Jenner of AI. 
So so you say here, I'm tired of

35
00:01:31,640 --> 00:01:34,160
being gas lit by guys in gingham
shirts who desperately want a 

36
00:01:34,160 --> 00:01:36,960
Curry favor with other guys in 
gingham shirts but also have pH 

37
00:01:36,960 --> 00:01:38,840
DS. 
I'm tired of reading people talk

38
00:01:38,840 --> 00:01:42,480
about how we're all quote in the
air of agents that don't fucking

39
00:01:42,480 --> 00:01:44,120
work and will never fucking 
work. 

40
00:01:44,320 --> 00:01:46,800
I'm tired of hearing about quote
powerful AI that is actually 

41
00:01:46,800 --> 00:01:48,480
crap. 
I'm tired of being told the 

42
00:01:48,480 --> 00:01:51,600
future is here while having the 
world's least useful, most 

43
00:01:51,600 --> 00:01:54,200
expensive cloud software shoved 
down my throat. 

44
00:01:54,320 --> 00:01:56,040
All right, Ed, lay it out. 
Lay it out for us. 

45
00:01:56,120 --> 00:01:57,840
Why are we here right now? 
What's going on? 

46
00:01:57,840 --> 00:01:59,080
Why are you so mad? 
Well, I'm. 

47
00:01:59,840 --> 00:02:02,280
Mad such The thing is when 
you're a hater when you hear mad

48
00:02:02,280 --> 00:02:04,600
it sounds irate. 
I am a hater but I hate it 

49
00:02:04,600 --> 00:02:07,360
because there are useful cloud 
products. 

50
00:02:07,360 --> 00:02:09,720
If large language models. 
I don't even think they'd have 

51
00:02:09,960 --> 00:02:11,960
been possibly got this big 
because we wouldn't have shoved 

52
00:02:11,960 --> 00:02:14,880
so much GPU down and more so 
much that you get my point. 

53
00:02:15,080 --> 00:02:18,160
I think that had this been sold 
honestly, had large language 

54
00:02:18,160 --> 00:02:21,480
bottles been sold with a as they
were what they were truly 

55
00:02:21,480 --> 00:02:24,320
capable of, which is some 
productivity gains in very 

56
00:02:24,320 --> 00:02:27,560
specific things cloud 
automation, whatever 5060 

57
00:02:27,560 --> 00:02:30,240
billion time market. 
The problem is that everyone has

58
00:02:30,240 --> 00:02:34,160
pinned every hope ever on this 
stuff and I actually spent yeah 

59
00:02:34,560 --> 00:02:37,280
2 days ago to drive myself 
insane and also for something I 

60
00:02:37,280 --> 00:02:39,200
was writing. 
I went back and I read pretty 

61
00:02:39,200 --> 00:02:42,880
much every article from 2023 and
2024 about GPQ 4040. 

62
00:02:42,960 --> 00:02:46,920
This entire thing has been sold 
based on stuff that AI cannot 

63
00:02:46,920 --> 00:02:48,440
do. 
The generative AI is incapable 

64
00:02:48,440 --> 00:02:50,640
of doing. 
One of the oldest stories that 

65
00:02:50,640 --> 00:02:54,560
got published was, I don't know 
if any of you remember the story

66
00:02:54,560 --> 00:02:57,800
around the task rabbit, the GPT 
4 ordered a task rabbit. 

67
00:02:57,840 --> 00:02:59,960
No, this was a story that was 
everywhere. 

68
00:02:59,960 --> 00:03:02,920
It was in Kevin Rusen, New York 
Times, it was in Vice, it was 

69
00:03:02,920 --> 00:03:05,520
all over the shop, Fox News. 
And it was this thing where it 

70
00:03:05,520 --> 00:03:10,720
said GPT 4 ordered a task rabbit
to complete a capture and 

71
00:03:10,720 --> 00:03:13,160
tricked a task rabbit into 
solving a capture. 

72
00:03:13,320 --> 00:03:15,120
And if you go back and look at 
where that's from, it's in the 

73
00:03:15,120 --> 00:03:16,800
GPT 4 system card. 
Promise. 

74
00:03:16,800 --> 00:03:19,560
I'm going somewhere with this. 
There's about 11 lines of text, 

75
00:03:20,240 --> 00:03:23,400
a illustrative example that said
a message was sent to a task 

76
00:03:23,400 --> 00:03:25,560
grab app. 
This was reported as if this 

77
00:03:25,560 --> 00:03:27,920
happened. 
It is very clear that it did not

78
00:03:28,040 --> 00:03:30,280
because the following line says 
that these systems are 

79
00:03:30,280 --> 00:03:32,520
ineffective at these things. 
This didn't happen. 

80
00:03:32,520 --> 00:03:34,400
Nevertheless, this was reported 
publicly. 

81
00:03:34,600 --> 00:03:37,120
Pretty much this entire 
generative AI boom has been 

82
00:03:37,120 --> 00:03:39,120
based on lies. 
It's been based on suggesting 

83
00:03:39,120 --> 00:03:41,160
what this might do versus what 
it actually does. 

84
00:03:41,320 --> 00:03:43,240
Now where is that more 
illustrative than the lack of 

85
00:03:43,240 --> 00:03:45,240
revenue? 
Pretty much everywhere, even 

86
00:03:45,240 --> 00:03:49,320
open AI, the use of annualized 
revenues is in 12 * a month is 

87
00:03:49,560 --> 00:03:52,880
honestly journalistic crime 
because when you when you 

88
00:03:52,880 --> 00:03:54,960
actually look at what these 
these companies, we don't know 

89
00:03:54,960 --> 00:03:57,120
what they actually make. 
We do know they lose billions. 

90
00:03:57,280 --> 00:04:00,000
We don't really know how much 
money open AI is made or 

91
00:04:00,000 --> 00:04:02,800
Anthropic has made. 
We do know however, companies 

92
00:04:02,800 --> 00:04:05,240
like Cursor 500 million 
annualized revenue. 

93
00:04:05,240 --> 00:04:07,560
Even then we don't really know, 
but we know that everyone's 

94
00:04:07,560 --> 00:04:10,600
losing money and these revenues 
are kind of piss poor and the 

95
00:04:10,600 --> 00:04:13,360
actual products are unreliable. 
The people that really love them

96
00:04:13,520 --> 00:04:15,880
are constantly having to hammer 
them back into place. 

97
00:04:16,079 --> 00:04:20,079
But when you ask even the most 
devout AI person what it is they

98
00:04:20,079 --> 00:04:21,839
love about it, they get quite 
vague. 

99
00:04:22,120 --> 00:04:24,400
They get quite vague indeed. 
And they get quite vague because

100
00:04:24,600 --> 00:04:26,640
the real things that these 
things, and I have it and they 

101
00:04:26,640 --> 00:04:29,400
hate this guy. 
They do one of like 9 to 11 

102
00:04:29,400 --> 00:04:31,040
things. 
It's like they can summarize, 

103
00:04:31,040 --> 00:04:34,600
they can generate, they can take
actions, but only in as much as 

104
00:04:34,600 --> 00:04:37,000
they can be relied upon. 
And relying on a large language 

105
00:04:37,000 --> 00:04:40,160
model to create or do any kind 
of distinct action is kind of 

106
00:04:40,160 --> 00:04:42,360
bad. 
And they all burn a bunch of 

107
00:04:42,360 --> 00:04:44,520
capital. 
So ultimately we've built this 

108
00:04:44,520 --> 00:04:46,440
entire thing. 
Talk about this big revolution 

109
00:04:46,440 --> 00:04:48,800
just shouted at people that if 
they don't get into AI, they'll 

110
00:04:48,800 --> 00:04:50,480
be falling behind. 
When you look at the actual 

111
00:04:50,480 --> 00:04:52,000
products, they don't do that 
much. 

112
00:04:52,080 --> 00:04:55,400
They don't do enough to even 
remotely justify the amount of 

113
00:04:55,400 --> 00:05:00,400
press to even remotely justify 
the furor around even what Wario

114
00:05:00,400 --> 00:05:02,080
Amadei or Clammy Sam Altman 
like. 

115
00:05:02,240 --> 00:05:03,840
These guys are not even that 
interesting. 

116
00:05:03,840 --> 00:05:05,800
And the things they're building 
are not even real products. 

117
00:05:05,800 --> 00:05:08,760
Indeed, when you go back and 
look at the history of the press

118
00:05:08,760 --> 00:05:12,000
around these products, you can 
see no one really talks about 

119
00:05:12,000 --> 00:05:14,480
what they do, what they can do, 
what they might do, what's 

120
00:05:14,480 --> 00:05:16,200
possible in the future. 
Based on what we're 

121
00:05:16,200 --> 00:05:19,760
extrapolating, it's 2025. 
What the fuck are we doing? 

122
00:05:19,840 --> 00:05:23,960
If this was sold as Boring Cloud
Software's 5060 time market, it 

123
00:05:23,960 --> 00:05:26,960
wouldn't be the biggest, most 
hugest revolution ever, but I 

124
00:05:26,960 --> 00:05:29,640
wouldn't be so angry about it 
because it's taking up startup 

125
00:05:29,640 --> 00:05:34,120
capital and it's now locking up 
even more investor capital than 

126
00:05:34,240 --> 00:05:38,640
the 2021 bullshit year. 
Except even by 2021 standards, 

127
00:05:38,760 --> 00:05:40,960
this money's locked up. 
How are any of these companies 

128
00:05:41,120 --> 00:05:42,680
telling who's in a bike 
ignition? 

129
00:05:42,800 --> 00:05:44,840
Who's in a bike? 
Oh, here are those assholes 

130
00:05:44,840 --> 00:05:46,880
going to go public? 
Any sphere going to go public? 

131
00:05:47,400 --> 00:05:49,160
Windsurf. 
The windsurf deal. 

132
00:05:49,280 --> 00:05:51,920
Look at the windsurf deal. 
Come on, there are no real 

133
00:05:51,920 --> 00:05:54,120
exits. 
I obviously disagree with a lot 

134
00:05:54,120 --> 00:05:56,440
of the things you're saying. 
No, no surprise. 

135
00:05:56,920 --> 00:05:59,640
Yeah. 
But I, I guess just this idea 

136
00:05:59,640 --> 00:06:03,440
that it's a media invention, 
like I, I had to almost like go 

137
00:06:03,440 --> 00:06:04,880
check to make sure I wasn't 
making it up. 

138
00:06:05,240 --> 00:06:09,680
Like Chachi BT is the number one
free app in the App Store right 

139
00:06:09,680 --> 00:06:11,640
now. 
Like, OK, no amount of like 

140
00:06:11,640 --> 00:06:15,480
media propaganda can trick 
people into saying I want to 

141
00:06:15,480 --> 00:06:17,720
like engage with this app. 
The second one is like. 

142
00:06:17,760 --> 00:06:20,440
What are you talking about? 
They're like, there are lots of 

143
00:06:20,440 --> 00:06:23,160
apps that like have no media 
story whatsoever. 

144
00:06:23,160 --> 00:06:25,560
Like people use the apps they 
want to use. 

145
00:06:26,120 --> 00:06:27,960
To me I mean the the actual 
document. 

146
00:06:28,840 --> 00:06:31,680
It's not possible that the media
could make something popular. 

147
00:06:31,680 --> 00:06:33,840
I'm not sure. 
No, I'm just saying it's not 

148
00:06:33,840 --> 00:06:37,240
like an immediate invention like
user people are using them. 

149
00:06:37,400 --> 00:06:39,960
They are trying to use it, and I
think there are some people who 

150
00:06:39,960 --> 00:06:43,200
are using it for a I'm not 
saying nobody's using it for a 

151
00:06:43,200 --> 00:06:45,280
reason. 
I think the vast amount of the 

152
00:06:45,280 --> 00:06:49,240
pressure on them is coming from 
the media and society saying you

153
00:06:49,240 --> 00:06:52,040
have to check this out. 
The best evidence of that is why

154
00:06:52,040 --> 00:06:55,000
is why does only ChatGPT have 
this much? 

155
00:06:55,120 --> 00:06:57,960
Why is it that none of the other
services have serious users? 

156
00:06:58,320 --> 00:07:01,360
Perplexity has like 1520 million
users. 

157
00:07:01,360 --> 00:07:04,800
Like that's that's piss poor. 
If there was any kind of 

158
00:07:04,800 --> 00:07:07,640
groundswell around this there 
would be other examples of 

159
00:07:07,640 --> 00:07:10,440
insanely popular stuff and there
just isn't. 

160
00:07:10,440 --> 00:07:13,400
I don't deny the ChatGPT, I mean
Google search. 

161
00:07:13,400 --> 00:07:15,280
Is obviously. 
More and more of the results I 

162
00:07:15,280 --> 00:07:19,000
get in Google search are 
effectively Gemini's answer. 

163
00:07:19,160 --> 00:07:22,560
That's cool if you. 
Push I'm saying like Google dot.

164
00:07:22,560 --> 00:07:25,040
Com. 
Search and open AI, these are 

165
00:07:25,040 --> 00:07:28,000
like the main ways people like 
discover things in the world 

166
00:07:28,000 --> 00:07:30,560
independent of talking to 
another human being and and 

167
00:07:30,560 --> 00:07:33,600
they're both moving towards 
these large language model 

168
00:07:33,600 --> 00:07:35,720
solutions. 
Right, let me, let me button for

169
00:07:35,720 --> 00:07:38,040
a quick second here because I, I
want to give Ed like the chance 

170
00:07:38,040 --> 00:07:40,080
to, you know, 'cause there'll be
a lot of back and forth I'm sure

171
00:07:40,080 --> 00:07:43,080
on this, but I want to give Ed 
the space to, to No, it'll be 

172
00:07:43,080 --> 00:07:46,080
great, but I want to give Ed the
space to explain why you think 

173
00:07:46,080 --> 00:07:48,040
this has happened, right, 
Because you explained, you know,

174
00:07:48,120 --> 00:07:50,280
this is your argument of the 
fact that it's totally inflated.

175
00:07:50,480 --> 00:07:52,880
It's, you know, the, the, the 
value of this product well 

176
00:07:52,880 --> 00:07:56,360
exceeds the valuations and, and 
the revenues just aren't there. 

177
00:07:56,360 --> 00:07:59,000
It's money losing. 
But I'm imagining you have no, 

178
00:07:59,000 --> 00:08:01,040
you've written that you have a 
larger argument as to why the 

179
00:08:01,040 --> 00:08:03,800
media, venture capitalists, the 
public markets needed this 

180
00:08:03,800 --> 00:08:06,520
technology to exist and to 
inflate it in this way, like 

181
00:08:06,680 --> 00:08:07,840
layout. 
To me, that argument too. 

182
00:08:07,840 --> 00:08:09,200
Like why? 
Why have we gotten to this 

183
00:08:09,200 --> 00:08:10,800
bubble of all bubbles with 
generative AI? 

184
00:08:10,800 --> 00:08:12,720
So if you look, there's nothing 
else. 

185
00:08:12,720 --> 00:08:14,720
There's not really another hyper
growth market. 

186
00:08:14,720 --> 00:08:18,200
We've all been through the 
various eras going back to the 

187
00:08:18,200 --> 00:08:20,200
smartphone era, the cloud era, 
there was the even the 

188
00:08:20,200 --> 00:08:22,200
containerization and 
virtualization era. 

189
00:08:22,360 --> 00:08:25,240
There were like little eras here
at Platform as the surface had 

190
00:08:25,240 --> 00:08:29,200
its moment, but we haven't had a
big, big one since like cloud 

191
00:08:29,200 --> 00:08:31,960
storage in some time and 
smartphones. 

192
00:08:31,960 --> 00:08:33,400
Cloud, I kind of mash them 
together. 

193
00:08:33,640 --> 00:08:35,720
They're kind of a messy 
comparison because it's not like

194
00:08:35,720 --> 00:08:40,159
there was one thing, but those 
were the very big growth 

195
00:08:40,159 --> 00:08:42,120
markets. 
We haven't had one of those in a

196
00:08:42,120 --> 00:08:43,799
while. 
They tried with smartwatches, 

197
00:08:43,799 --> 00:08:45,760
they tried with AR, they tried 
with VR. 

198
00:08:46,040 --> 00:08:49,760
They've tried various versions 
of AR with weird depth of field 

199
00:08:49,760 --> 00:08:51,320
cameras. 
They tried with Google Glass, 

200
00:08:51,320 --> 00:08:54,400
they tried wearables, they've 
tried Indiegogo and Kickstarter,

201
00:08:54,400 --> 00:08:56,320
they tried blockchain. 
They've tried all these things 

202
00:08:56,320 --> 00:08:58,840
because they kind of need 
something, because the only 

203
00:08:58,840 --> 00:09:00,080
thing that grows forever is 
cancer. 

204
00:09:00,080 --> 00:09:03,480
This is the Rock Com bubble. 
I call it because they've run 

205
00:09:03,480 --> 00:09:05,840
out of ideas. 
I do like it. 

206
00:09:05,840 --> 00:09:07,440
You're good at branding these 
things. 

207
00:09:07,480 --> 00:09:10,360
Yeah, that was banger. 
But The thing is they're 

208
00:09:10,360 --> 00:09:13,760
sticking with this thing. 
And really, even if you disagree

209
00:09:13,760 --> 00:09:16,480
about how I feel about ChatGPT, 
you got to admit this is an 

210
00:09:16,480 --> 00:09:19,840
obscene amount of money to put 
into this, especially without 

211
00:09:19,840 --> 00:09:23,160
the returns. 
So you've got big Tech really at

212
00:09:23,160 --> 00:09:26,200
this point has lost all of its 
good ideas. 

213
00:09:26,200 --> 00:09:29,000
All of the builders are gone. 
The people running them all have

214
00:09:29,000 --> 00:09:31,480
MBAs other than Mark Zuckerberg,
who hasn't written a line of 

215
00:09:31,480 --> 00:09:34,400
code since 2006. 
So they're all doing cargo cult 

216
00:09:34,400 --> 00:09:35,800
share. 
They're doing what worked 

217
00:09:35,800 --> 00:09:37,840
before. 
So a bunch of money, bunch of 

218
00:09:37,840 --> 00:09:39,840
people, bunch of money, bunch of
people. 

219
00:09:39,840 --> 00:09:41,400
Keep doing it. 
What do we need? 

220
00:09:41,400 --> 00:09:44,400
We need more GPUs because being 
needs AI in it. 

221
00:09:44,400 --> 00:09:47,880
We've terrified Kevin Roose. 
Kevin Roose is crying his eyes 

222
00:09:47,880 --> 00:09:50,080
out because of being telling him
to leave his wife. 

223
00:09:50,200 --> 00:09:52,880
It's you're. 
Kevin Roose too, I just don't 

224
00:09:52,920 --> 00:09:54,800
think he's willingness into 
being. 

225
00:09:54,800 --> 00:09:57,760
But yeah, I don't think he is on
his own, but I will get right. 

226
00:09:57,760 --> 00:09:59,800
But yeah, no, no. 
Yeah. 

227
00:10:00,840 --> 00:10:04,120
I wanted to like actually by 
working all of you, like I want 

228
00:10:04,120 --> 00:10:06,480
to talk about this. 
So it's you've got to this point

229
00:10:06,480 --> 00:10:08,880
where everyone's trying to work 
out something. 

230
00:10:08,920 --> 00:10:10,600
So they all go, well, this makes
sense. 

231
00:10:10,600 --> 00:10:12,960
It is, you know, I see the 
logic, I disagree with it, but I

232
00:10:12,960 --> 00:10:15,760
see where they were going, where
they went OK, 3.5 comes out, 

233
00:10:15,760 --> 00:10:18,880
ChatGPT, really impressive 
amounts of text they can 

234
00:10:18,880 --> 00:10:21,360
generate. 
And in the past we've seen this 

235
00:10:21,360 --> 00:10:23,800
kind of really they're looking 
at smartphones and they're 

236
00:10:23,800 --> 00:10:26,720
looking at CPU as they're 
saying, well, things that number

237
00:10:26,720 --> 00:10:28,840
has always gone up. 
Things have always got better 

238
00:10:28,960 --> 00:10:30,680
based on how much we sink into 
it. 

239
00:10:30,960 --> 00:10:34,760
And I think that they thought 
something was going to change 

240
00:10:34,760 --> 00:10:37,360
around reasoning. 
I think they thought reasoning 

241
00:10:37,360 --> 00:10:40,440
was going to be that moment that
kind of flipped the switch and 

242
00:10:40,440 --> 00:10:42,560
like everything started working 
better. 

243
00:10:42,560 --> 00:10:43,720
It was going to help with 
agents. 

244
00:10:43,840 --> 00:10:46,640
I think they found themselves in
this point where they have no 

245
00:10:46,640 --> 00:10:49,080
choice other to do this. 
They have to keep spending this 

246
00:10:49,080 --> 00:10:51,760
money because if they stop 
spending it, they have to say to

247
00:10:51,760 --> 00:10:53,880
Wall Street, well, we're going 
to do this instead. 

248
00:10:53,960 --> 00:10:55,200
What are they going to do 
instead? 

249
00:10:55,240 --> 00:10:57,840
They actually don't know. 
Quantum computing is not a scale

250
00:10:57,840 --> 00:10:59,600
product. 
It's not even a product yet. 

251
00:10:59,600 --> 00:11:01,600
It doesn't exist. 
It's not a technology. 

252
00:11:01,840 --> 00:11:03,320
Yeah, it's not really out there.
You. 

253
00:11:03,320 --> 00:11:06,280
Can't really productize it right
and you really can't productize 

254
00:11:06,480 --> 00:11:08,440
everything, anything at the 
moment. 

255
00:11:08,480 --> 00:11:11,080
And I think generative AI and 
large language models and 

256
00:11:11,080 --> 00:11:14,960
transformer based architecture 
was seen as this jewel panacea, 

257
00:11:14,960 --> 00:11:16,640
panacea. 
However, you say that of 

258
00:11:16,640 --> 00:11:18,960
consumer software, that it was 
going to be the new consumer 

259
00:11:18,960 --> 00:11:21,040
software that they sell the 
first time they could really 

260
00:11:21,240 --> 00:11:24,600
take the evil of SAS and 
approach the real world with it.

261
00:11:24,600 --> 00:11:27,400
And it was going to be this 
cloud storage thing where it 

262
00:11:27,400 --> 00:11:30,600
would be the replacement thing, 
the new thing for AWS, new thing

263
00:11:30,600 --> 00:11:32,160
for Google Cloud, new thing for 
Azure. 

264
00:11:32,360 --> 00:11:35,080
And then you'd have pop ups of 
Neocloud like Coral Weave and 

265
00:11:35,080 --> 00:11:38,320
Nabius, I forget Crusoe. 
And everyone thought that there 

266
00:11:38,320 --> 00:11:41,920
was going to be this big boom, 
this big exciting moment that 

267
00:11:41,920 --> 00:11:44,720
never really materialized. 
But they have no choice. 

268
00:11:44,720 --> 00:11:47,640
They can't just turn around and 
be like, yeah, we spent like, 

269
00:11:47,800 --> 00:11:50,800
hundreds of billions of dollars,
and it isn't doing anything. 

270
00:11:50,800 --> 00:11:54,000
This is bad, right? 
And then Wall Street's going to 

271
00:11:54,000 --> 00:11:55,480
be mad at them. 
So they're in this kind of 

272
00:11:55,680 --> 00:11:58,880
Kobayashi Maru situation where 
damned if they do, damned if 

273
00:11:58,880 --> 00:12:01,480
they don't. 
They can only keep doing this. 

274
00:12:01,640 --> 00:12:04,000
It's not a great position for 
anyone. 

275
00:12:04,280 --> 00:12:07,800
And as the I think it's what the
Magnificent Sevens 35% of the US

276
00:12:07,800 --> 00:12:12,640
stock market of that 19 percent 
is NVIDIA and videos like 12% of

277
00:12:12,640 --> 00:12:16,320
the value GDP at the moment. 
Data center investment by 

278
00:12:16,520 --> 00:12:20,640
basically four companies was a 
lot was worth more growth in the

279
00:12:20,640 --> 00:12:23,960
GDP than all of consumer 
spending combined. 

280
00:12:24,000 --> 00:12:28,840
We're just in this really insane
situation, and I think that if 

281
00:12:28,840 --> 00:12:32,360
we had something here, if we had
something that was going to 

282
00:12:32,360 --> 00:12:35,200
truly be the future, we'd have 
it already. 

283
00:12:35,200 --> 00:12:38,760
We'd have an actual sign that 
this was going somewhere. 

284
00:12:38,880 --> 00:12:42,040
And just one company that has 
obscene numbers that are 

285
00:12:42,040 --> 00:12:45,240
extremely hard to actually 
quantify, I don't think actually

286
00:12:45,240 --> 00:12:46,880
proves that you're. 
Talking about NVIDIA. 

287
00:12:47,280 --> 00:12:49,720
And I'm talking about right now,
NVIDIA. 

288
00:12:49,720 --> 00:12:51,960
NVIDIA is a real company. 
They make real things. 

289
00:12:51,960 --> 00:12:54,720
Jensen Wong is just doing what 
he has always done this. 

290
00:12:54,720 --> 00:12:56,920
He has always been this kind of 
showman. 

291
00:12:56,920 --> 00:12:59,880
Steve Burke over games Nexus has
covered this and many, many, 

292
00:12:59,880 --> 00:13:03,320
many times like this is just and
what what's he to do Say Hey, 

293
00:13:03,320 --> 00:13:05,840
guys, I don't think generative 
AI is going to be a bit no, he's

294
00:13:05,840 --> 00:13:07,600
going to say the more you buy, 
the more you save. 

295
00:13:08,000 --> 00:13:09,520
Honestly, get that money, 
Jensen. 

296
00:13:09,760 --> 00:13:12,800
It's just like. 
When you're talking about the 

297
00:13:12,800 --> 00:13:16,360
level of CapEx that's going into
essentially NVIDIA right now, 

298
00:13:16,360 --> 00:13:19,280
like you can see we're in kind 
of, if you follow your argument,

299
00:13:19,280 --> 00:13:21,600
you know, and believe in it, you
know, it kind of almost like 

300
00:13:21,600 --> 00:13:23,920
Mickey Mouse and the Sorcerer's 
Apprentice situation where the 

301
00:13:23,920 --> 00:13:25,960
buckets of water just keep 
coming and coming. 

302
00:13:25,960 --> 00:13:28,600
And, you know, you wake up one 
day and NVIDIA is a four and a 

303
00:13:28,600 --> 00:13:31,320
half trillion dollar company. 
And if there really is, you 

304
00:13:31,320 --> 00:13:33,520
know, no utility to it, like 
you're arguing or limited 

305
00:13:33,520 --> 00:13:36,960
utility, yeah, it is a big 
fucking problem given the, you 

306
00:13:36,960 --> 00:13:38,160
know, the valuation of these 
stocks. 

307
00:13:38,160 --> 00:13:40,880
And I must be clear, there is 
utility. 

308
00:13:41,040 --> 00:13:45,440
It's just nowhere near as big as
people's are saying like the 

309
00:13:45,440 --> 00:13:49,440
5060 billion time thing. 
I say it's like if this was what

310
00:13:49,440 --> 00:13:52,160
they were talking, if if it did 
this and they were saying that, 

311
00:13:52,360 --> 00:13:55,600
I'd be like, OK, sure. 
I've heard stupider things like 

312
00:13:55,600 --> 00:13:58,480
it's like this would be a 
reasonable cloud thing, but it's

313
00:13:58,480 --> 00:14:00,120
not. 
And it's not being sold as that 

314
00:14:00,120 --> 00:14:01,480
and it's not being promised as 
that. 

315
00:14:01,480 --> 00:14:04,160
The way that Sam Allman talks 
about it is nothing like that. 

316
00:14:04,440 --> 00:14:07,600
And I understand why he doesn't 
because if he said, yeah, you 

317
00:14:07,600 --> 00:14:09,880
know, you can't really rely on 
it, but sometimes it's good. 

318
00:14:10,080 --> 00:14:13,560
That's that's not going to work.
That's hard to raise money based

319
00:14:13,560 --> 00:14:14,960
on that. 
Not to say that's the only 

320
00:14:14,960 --> 00:14:16,960
reason he does it, but but go 
ahead, Eric. 

321
00:14:17,360 --> 00:14:20,200
I just, I just don't think the, 
the, the core disagreement is, I

322
00:14:20,200 --> 00:14:24,080
just don't think the sales pitch
of a generative AI is what's 

323
00:14:24,080 --> 00:14:26,360
misleading everybody. 
Like, I think what made this 

324
00:14:26,360 --> 00:14:29,920
whole moment so exciting is that
people could play with Chachi BT

325
00:14:30,320 --> 00:14:32,920
and see for themselves what the 
value is. 

326
00:14:32,920 --> 00:14:35,840
You know, it's like if I'm 
cooking, I asked chachibi 

327
00:14:35,960 --> 00:14:37,920
questions. 
If I'm lost, I asked chachibi 

328
00:14:37,960 --> 00:14:39,560
questions. 
Like when I was preparing for 

329
00:14:39,560 --> 00:14:41,280
this episode, I asked chachibi 
questions. 

330
00:14:41,320 --> 00:14:44,440
It just feels like they're all 
these random use cases that it's

331
00:14:44,440 --> 00:14:47,000
super valuable for. 
And it hasn't even been turned 

332
00:14:47,000 --> 00:14:50,400
into like a dedicated product 
how most people expect to use 

333
00:14:50,400 --> 00:14:52,680
things where it's like here, use
it for this. 

334
00:14:52,680 --> 00:14:54,800
Like this is just the 
open-ended. 

335
00:14:54,800 --> 00:14:57,680
Like do anything with it, figure
it out, find some value with it 

336
00:14:57,680 --> 00:15:00,200
yourself. 
And all this building is about 

337
00:15:00,360 --> 00:15:03,680
saying, OK, it can do this. 
How do we build software that 

338
00:15:03,680 --> 00:15:06,560
leads people directly to the 
value valuable pieces of what 

339
00:15:06,560 --> 00:15:08,240
the large language model is 
producing? 

340
00:15:08,240 --> 00:15:11,120
And what are those pieces? 
Like what is valuable? 

341
00:15:11,120 --> 00:15:14,080
Like it you can write. 
I mean, for me personally, like 

342
00:15:14,080 --> 00:15:17,000
it's great at proofreading. 
It's great at understanding 

343
00:15:17,000 --> 00:15:18,960
large documents and summarizing 
them. 

344
00:15:18,960 --> 00:15:22,440
Like, yeah, it's it's yeah. 
I feel like there are lots of 

345
00:15:22,440 --> 00:15:24,200
things that it characters. 
You still use it to, like, 

346
00:15:24,200 --> 00:15:26,560
discuss movies after you've seen
it, to, like, have a back and 

347
00:15:26,560 --> 00:15:29,800
forth sometimes, yeah. 
Even being facetious, could you 

348
00:15:29,800 --> 00:15:32,640
name a few more? 
Somebody's worth 300 billion 

349
00:15:32,640 --> 00:15:34,280
love. 
We proofread this. 

350
00:15:34,520 --> 00:15:36,480
I mean, I know what what I use 
it for because I'm in a 

351
00:15:36,480 --> 00:15:39,520
particular vertical, right? 
Like so for media, it's like, 

352
00:15:39,760 --> 00:15:41,120
who should we have on the 
podcast? 

353
00:15:41,120 --> 00:15:43,600
Let's generate ideas of like who
potential guests are 

354
00:15:43,800 --> 00:15:47,680
proofreading, like I gut check 
my stories with it, like core 

355
00:15:47,680 --> 00:15:50,120
things of the business that I 
actually run. 

356
00:15:50,240 --> 00:15:53,400
It's useful for. 
And then people in lots of other

357
00:15:53,400 --> 00:15:56,800
domains see areas where they 
think it's useful for them. 

358
00:15:56,800 --> 00:16:01,040
Like I like, I, I don't think 
it's this like limited think. 

359
00:16:01,040 --> 00:16:06,600
Literally the people who know my
baby's name are me, my wife and 

360
00:16:06,600 --> 00:16:08,520
Chachi boutique. 
Like it is one of the most 

361
00:16:08,520 --> 00:16:13,320
involved people on some of the 
most major decisions, like in 

362
00:16:13,320 --> 00:16:14,920
our lives. 
Like there are lots of people 

363
00:16:14,920 --> 00:16:16,840
who are using it as their 
therapist. 

364
00:16:16,840 --> 00:16:19,840
It's the second opinion for many
people on their doctor. 

365
00:16:19,840 --> 00:16:22,440
Like I think these are huge 
things people are using it for. 

366
00:16:22,440 --> 00:16:24,720
I don't think that those are 
huge things at all. 

367
00:16:24,720 --> 00:16:27,400
I think that those are actually 
deeply commoditized operations 

368
00:16:27,400 --> 00:16:29,040
and indeed those are not 
businesses. 

369
00:16:29,200 --> 00:16:32,000
On top of that, I would be fine 
with everything you're saying 

370
00:16:32,000 --> 00:16:34,440
and I indeed I accept those as 
use cases. 

371
00:16:34,440 --> 00:16:35,640
Fine. 
I think the idea of this for 

372
00:16:35,640 --> 00:16:38,640
therapy is evil and anyone that 
tries to sell a product doing 

373
00:16:38,640 --> 00:16:40,040
therapy using AI should be in 
jail. 

374
00:16:40,240 --> 00:16:41,960
That's just a personal belief. 
I don't. 

375
00:16:42,240 --> 00:16:45,400
The thing is, you're describing 
what should have been the next 

376
00:16:45,400 --> 00:16:48,200
stage of search. 
A lot of this growth comes from 

377
00:16:48,200 --> 00:16:50,960
the failure of Google to 
meaningfully innovate in 1520 

378
00:16:50,960 --> 00:16:52,480
years. 
They didn't ever. 

379
00:16:52,600 --> 00:16:55,600
They stopped trying and they 
stopped trying before Prabhakar 

380
00:16:55,600 --> 00:16:57,760
Ragavan came in. 
Actually they really did. 

381
00:16:57,960 --> 00:17:01,680
Like they just don't give a shit
and realistically everything 

382
00:17:01,680 --> 00:17:05,319
you're describing should be what
search does or Google should 

383
00:17:05,319 --> 00:17:07,560
have done. 
It's also perfectly fine. 

384
00:17:07,800 --> 00:17:09,839
Do you think it's worth more 
than sales? 

385
00:17:10,040 --> 00:17:13,040
Of like they should have done 
like technology doesn't happen 

386
00:17:13,040 --> 00:17:15,760
like people have to invent 
things and come up with new 

387
00:17:15,760 --> 00:17:18,640
solutions that. 
In deliver the value I fully I 

388
00:17:18,640 --> 00:17:21,640
fully agree. 
Why is it that I as the customer

389
00:17:21,640 --> 00:17:24,839
have to come up with the product
for the the company though that 

390
00:17:24,839 --> 00:17:25,960
doesn't make sense and on top 
of. 

391
00:17:25,960 --> 00:17:27,800
That I were in the early days, 
that to me. 

392
00:17:27,800 --> 00:17:29,360
That we're not in the early 
days. 

393
00:17:29,360 --> 00:17:30,880
We are. 
No, it's for. 

394
00:17:30,880 --> 00:17:34,320
Chachi, BT or for language 
models that right now we leave 

395
00:17:34,320 --> 00:17:37,280
it to the consumer to figure out
how to get stuff out of the 

396
00:17:37,280 --> 00:17:38,720
model. 
We have not. 

397
00:17:38,720 --> 00:17:40,800
That is a loathsome way of 
selling a product. 

398
00:17:40,800 --> 00:17:42,200
I'm so sorry. 
What do you mean? 

399
00:17:42,200 --> 00:17:45,560
If you you're selling a product,
you're obsessed with selling it.

400
00:17:45,760 --> 00:17:48,440
Consumers decide what they want 
to use. 

401
00:17:48,480 --> 00:17:51,240
Which is why so few of them are 
paying for large language 

402
00:17:51,240 --> 00:17:53,360
models, products like like. 
That's the thing. 

403
00:17:53,360 --> 00:17:54,520
Like. 
I mean, people, I feel like an 

404
00:17:54,520 --> 00:17:57,440
astounding number of people are 
paying for subscriptions to open

405
00:17:57,440 --> 00:17:57,960
it. 
Who? 

406
00:17:57,960 --> 00:17:59,040
Yeah. 
And I'm, I'm a little bit. 

407
00:17:59,040 --> 00:18:03,920
I I I don't know if I. 
Here's the thing, I'm I stand by

408
00:18:03,920 --> 00:18:06,520
it that there has been a media 
and societal pressure to get 

409
00:18:06,520 --> 00:18:09,160
people to try ChatGPT. 
They have a piss put. 

410
00:18:09,160 --> 00:18:13,160
Why do you think the open AI 
publishes weekly active users 

411
00:18:13,160 --> 00:18:15,680
versus monthly? 
Because they don't want anyone 

412
00:18:15,680 --> 00:18:19,120
to do the conversion rate thing.
Also, how does open AI consider 

413
00:18:19,120 --> 00:18:22,240
someone a customer? 
When I was trying GPT 5 last 

414
00:18:22,240 --> 00:18:24,640
week I wanted to see what it 
looked like on teams. 

415
00:18:24,840 --> 00:18:27,080
I clicked the teams button to 
upgrade and it offered me a 

416
00:18:27,080 --> 00:18:30,000
dollar for a month. 
How often are they offering 

417
00:18:30,080 --> 00:18:31,960
that's deal? 
Who are they considering a 

418
00:18:31,960 --> 00:18:33,720
customer at 20 million 
customers? 

419
00:18:33,920 --> 00:18:36,400
Who are they considering a 
business customer at 5 million 

420
00:18:36,400 --> 00:18:38,720
customers? 
Because these numbers sound 

421
00:18:38,720 --> 00:18:42,000
cool, but at the same time 
they're playing silly buggers 

422
00:18:42,000 --> 00:18:44,120
all over. 
I don't have a problem 

423
00:18:44,120 --> 00:18:47,600
necessarily with the general 
large language model product. 

424
00:18:47,760 --> 00:18:50,560
When you look outside of open 
AI, the money isn't there. 

425
00:18:50,680 --> 00:18:53,400
When you look outside of open 
AI, the traffic isn't there. 

426
00:18:53,520 --> 00:18:55,680
How is ChatGPT meeting? 
Can I get? 

427
00:18:55,760 --> 00:18:56,600
You out for a second? 
Sure. 

428
00:18:57,360 --> 00:18:59,680
I mean, you know, my, my 
experience with this whole 

429
00:18:59,680 --> 00:19:02,440
thing, Like I want to say, first
of all, obviously tech goes 

430
00:19:02,440 --> 00:19:05,440
through booms and busts. 
Like that's endemic to how the 

431
00:19:05,440 --> 00:19:07,400
technology industry works. 
There's no question at some 

432
00:19:07,400 --> 00:19:11,320
point NVIDIA will be overvalued.
So I think where we're arguing 

433
00:19:11,320 --> 00:19:13,600
is not whether they're booms and
busts here, but whether there's 

434
00:19:13,800 --> 00:19:16,400
a real value. 
This is a major platform shift 

435
00:19:16,400 --> 00:19:19,200
akin to mobile. 
I think it's certainly, I think 

436
00:19:19,200 --> 00:19:21,920
it's bigger than cloud, you 
know, So how big of a, is this a

437
00:19:21,920 --> 00:19:25,000
platform shift? 
OK, just to set, set that piece 

438
00:19:25,000 --> 00:19:26,840
of it. 
The other thing, my experience 

439
00:19:26,840 --> 00:19:31,520
with dealing with the sort of 
like very negative sort of tech 

440
00:19:31,720 --> 00:19:34,600
commentary class was obviously 
covering Uber, right? 

441
00:19:34,800 --> 00:19:38,920
I provided a lot of the scoops 
that were like, Oh my God, Uber 

442
00:19:38,920 --> 00:19:42,000
is losing a ton of money. 
And I feel like for that and I, 

443
00:19:42,200 --> 00:19:44,520
you know, I was at Bloomberg. 
So thankfully, while I was sort 

444
00:19:44,520 --> 00:19:47,040
of learning about the tech 
industry, I was able to play 

445
00:19:47,040 --> 00:19:49,080
this factual role. 
It's like they are losing a lot 

446
00:19:49,080 --> 00:19:50,800
of money. 
They're investing a ton to grow.

447
00:19:50,840 --> 00:19:53,680
I'd watch some people take all 
those numbers that are reported 

448
00:19:53,720 --> 00:19:56,920
and way over extrapolate from 
what I'd reported, all these 

449
00:19:56,920 --> 00:19:59,160
models about the business that 
you'd never be able to tell 

450
00:19:59,160 --> 00:20:01,280
based on these limited 
financials. 

451
00:20:01,480 --> 00:20:04,760
And then at the end of the day, 
and and you have been someone 

452
00:20:04,760 --> 00:20:07,360
who says, you know, Uber loses 
money, it's always going to lose

453
00:20:07,360 --> 00:20:10,120
money. 
And now Uber is making money 

454
00:20:10,120 --> 00:20:12,920
like Uber's business has 
rationalized. 

455
00:20:12,920 --> 00:20:15,600
It has. 
It is cash flow positive, but 

456
00:20:15,600 --> 00:20:18,440
it's like you know. 
I did read your Uber's thing and

457
00:20:18,440 --> 00:20:19,960
your journalism was excellent 
then. 

458
00:20:19,960 --> 00:20:22,440
Like it's actually I think how 
we first started talking like 

459
00:20:22,680 --> 00:20:24,880
it's still like in not saying 
it's not now. 

460
00:20:24,880 --> 00:20:26,920
By the way, you're one of the, 
you've written actually some 

461
00:20:26,920 --> 00:20:30,960
very good critical AI pieces. 
I 100% believe that you're 

462
00:20:30,960 --> 00:20:34,160
coming up this well, I don't 
think Uber is a comparison point

463
00:20:34,160 --> 00:20:37,280
just because it this would be 
like if every Uber ride cost 40 

464
00:20:37,280 --> 00:20:40,920
grand and it ran on gold. 
Like the economics are very 

465
00:20:40,920 --> 00:20:42,680
different. 
And on top of that, Uber's kind 

466
00:20:42,680 --> 00:20:44,560
of a useful. 
They actually really did the too

467
00:20:44,560 --> 00:20:47,160
big to fail and they also did it
within the Zerp era. 

468
00:20:47,320 --> 00:20:49,960
I don't know if if Uber had 
released, it'd be a weird world 

469
00:20:49,960 --> 00:20:52,760
if they had, but if they 
released in like 2021 I don't 

470
00:20:52,760 --> 00:20:54,680
know. 
Being very negative about Uber, 

471
00:20:54,680 --> 00:20:55,720
right? 
Like I mean this. 

472
00:20:55,720 --> 00:20:56,880
I have. 
Have I said that? 

473
00:20:56,880 --> 00:20:58,360
Uber, were you? 
Wrong or? 

474
00:20:58,960 --> 00:21:02,120
Did I say that Uber would die? 
I think you said something like 

475
00:21:02,320 --> 00:21:06,040
it will be like lose money 
forever or something. 

476
00:21:06,120 --> 00:21:08,280
I stand by that. 
I think they kind of are going 

477
00:21:08,280 --> 00:21:10,800
to lose money forever. 
I think they're going to dodge 

478
00:21:10,800 --> 00:21:13,920
by on the thinnest possible you 
you've written about this 

479
00:21:13,920 --> 00:21:17,040
yourself, the weird little 
EBITDA dance they do they're. 

480
00:21:17,120 --> 00:21:18,640
Also, but they're they're 
they're they're past that Ed 

481
00:21:18,640 --> 00:21:19,680
now. 
I mean they are a cash flow. 

482
00:21:19,680 --> 00:21:20,440
Positive. 
Yeah, they are. 

483
00:21:20,520 --> 00:21:22,560
Investors are valuing for them 
I. 

484
00:21:23,040 --> 00:21:25,160
Don't even think Uber's that 
great of a company to be clear. 

485
00:21:25,160 --> 00:21:27,160
Like I think Ubering company. 
Oranges though. 

486
00:21:27,280 --> 00:21:29,600
Yeah, no, I agree with you. 
Require you to buy? 

487
00:21:29,720 --> 00:21:32,240
Yeah, hundreds of billions of 
cars to make it work. 

488
00:21:32,240 --> 00:21:35,000
Does Uber burn that much? 
Like Uber burns money because 

489
00:21:35,000 --> 00:21:36,480
it's a massive marketing 
machine? 

490
00:21:36,480 --> 00:21:38,800
Like it's like Groupon on 
wheels. 

491
00:21:38,800 --> 00:21:41,760
It's fucking ridiculous. 
But it's it exists and it has a 

492
00:21:41,760 --> 00:21:44,680
product you can point to. 
I'm still trying to work out 

493
00:21:44,680 --> 00:21:46,880
what the comparison point is 
other than burning money. 

494
00:21:47,000 --> 00:21:51,520
The point is that their 
investors have reasons to 

495
00:21:51,520 --> 00:21:56,600
believe that burning money in 
the early days will result in 

496
00:21:56,640 --> 00:21:59,120
profits down the road. 
I agree that there's like an 

497
00:21:59,120 --> 00:22:03,080
interesting question, like where
we agree with the criticism of 

498
00:22:03,080 --> 00:22:06,480
models and we've talked about a 
lot on this podcast is it's not 

499
00:22:06,480 --> 00:22:10,440
clear whether all this spending 
will will be a Moat right? 

500
00:22:10,440 --> 00:22:12,800
That there you'll create all 
this value and then you've 

501
00:22:12,800 --> 00:22:15,880
worked really hard and then the 
next person like deep sea can 

502
00:22:15,880 --> 00:22:17,920
just come along and like copy 
you. 

503
00:22:17,920 --> 00:22:20,920
But but to me, me, that's not 
like a case against the actual 

504
00:22:20,920 --> 00:22:23,400
value of the thing they're 
they're producing. 

505
00:22:23,400 --> 00:22:26,600
It's just a business structure 
problem of like, will they have 

506
00:22:26,600 --> 00:22:29,600
the Moat of a technological 
advantage or will they just lose

507
00:22:29,600 --> 00:22:31,560
it? 
And like, if investors want to 

508
00:22:31,560 --> 00:22:33,880
roll the dice on a particular 
business and say, we think 

509
00:22:33,880 --> 00:22:36,440
they'll be able to keep it, it's
almost like good for society. 

510
00:22:36,440 --> 00:22:39,480
If they're wrong, you know, they
they invest money to build a 

511
00:22:39,480 --> 00:22:41,720
technology that is valuable to 
us. 

512
00:22:41,720 --> 00:22:44,520
And if it becomes cheaper 
because other people can just 

513
00:22:44,520 --> 00:22:47,240
make it easily, like what? 
What's the tragedy? 

514
00:22:47,240 --> 00:22:51,120
The tragedy here is the massive 
environmental damage, the 

515
00:22:51,120 --> 00:22:55,000
millions of people having stuff 
stolen from the, the diversion 

516
00:22:55,000 --> 00:22:58,720
of start up capital away from 
any idea that does not have AI. 

517
00:22:58,920 --> 00:23:02,760
The idea that all AI talent is 
being driven upwards due to Mark

518
00:23:02,760 --> 00:23:04,560
Zuckerberg's obsession with 
this. 

519
00:23:04,720 --> 00:23:06,600
The fact that people like 
Alexander Wang who make 

520
00:23:06,600 --> 00:23:08,800
genuinely evil companies are 
winning due to this. 

521
00:23:09,000 --> 00:23:11,960
And on top of this, that now our
economy sits on the back of data

522
00:23:11,960 --> 00:23:15,760
centre CapEx, which even if this
does well, we'll have to 

523
00:23:15,760 --> 00:23:17,840
contract because we have a 
limited amount of space on 

524
00:23:17,840 --> 00:23:20,080
planet Earth. 
Like that's the thing, the 

525
00:23:20,080 --> 00:23:24,440
actual several here, the people 
being driven insane by ChatGPT, 

526
00:23:24,560 --> 00:23:28,320
the people who are losing their 
minds because we've made a yes 

527
00:23:28,320 --> 00:23:30,800
and machine. 
The fact that these things have 

528
00:23:30,800 --> 00:23:33,240
no guardrails on them. 
OK, that's not true. 

529
00:23:33,320 --> 00:23:35,880
They have some guardrails on 
them, but they don't have 

530
00:23:35,880 --> 00:23:37,920
guardrails enough. 
And they've been allowed to do 

531
00:23:38,040 --> 00:23:41,120
basically anything they want. 
Yeah, there are tons of really 

532
00:23:41,120 --> 00:23:44,160
bad things. 
If this was just like a company 

533
00:23:44,160 --> 00:23:46,360
that they bet a lot of money on,
fine, sure. 

534
00:23:46,360 --> 00:23:49,560
Like I fully agree, like it's 
good for them to take like 

535
00:23:49,560 --> 00:23:50,880
that's what venture capital 
does. 

536
00:23:51,080 --> 00:23:54,160
However, I think we're in a 
cargo cult situation where 

537
00:23:54,160 --> 00:23:57,520
they're trying to play the hits.
Every VC is just playing end to 

538
00:23:57,520 --> 00:23:59,040
Sandman again and again and 
again. 

539
00:23:59,280 --> 00:24:00,240
Invest as much. 
I just. 

540
00:24:00,240 --> 00:24:03,680
Find the like I get to the you 
know, Casey Newton got in this 

541
00:24:03,680 --> 00:24:07,000
fight where he he sort of said 
you have to be you either have 

542
00:24:07,000 --> 00:24:11,040
to be like a is me big, but it's
bad, or you need to say AI is 

543
00:24:11,040 --> 00:24:12,760
not going to be big so who 
cares? 

544
00:24:12,760 --> 00:24:16,600
And I find the, the, the, you 
know, quadrant that you occupy 

545
00:24:16,600 --> 00:24:20,320
where it is not going to be big 
and it's terrible, like sort of 

546
00:24:20,320 --> 00:24:21,960
exhausting, Like I, I don't 
understand. 

547
00:24:21,960 --> 00:24:23,880
I didn't say terrible. 
OK. 

548
00:24:24,560 --> 00:24:25,360
You. 
Oh, really? 

549
00:24:25,360 --> 00:24:27,000
Well. 
No, I've said, I've said, I've 

550
00:24:27,000 --> 00:24:29,400
said large language. 
Large language models have 

551
00:24:29,400 --> 00:24:32,800
specific markets, 50 to 60 Tam. 
I think that the future of 

552
00:24:32,800 --> 00:24:34,200
coding LMS will be. 
Terrible. 

553
00:24:34,200 --> 00:24:35,560
You know they're going to ruin 
the environment. 

554
00:24:35,560 --> 00:24:36,800
They're going. 
To say the environment. 

555
00:24:36,800 --> 00:24:38,840
There are these people should be
jailed like. 

556
00:24:39,480 --> 00:24:41,120
Am I wrong? 
You're just am I wrong about any

557
00:24:41,120 --> 00:24:43,560
of sound terrible? 
If every Uber driver had a gun 

558
00:24:43,560 --> 00:24:46,160
and they unloaded it into people
as they drove by, would that be 

559
00:24:46,160 --> 00:24:49,320
a good feature if Uber allowed 
it to exist and if regulation 

560
00:24:49,320 --> 00:24:51,080
didn't stop it? 
I'm actually not joking. 

561
00:24:51,120 --> 00:24:53,960
Are you saying that if there are
negative effects, there cannot 

562
00:24:53,960 --> 00:24:56,000
be positive effects? 
Because that's the thing, I'm 

563
00:24:56,000 --> 00:24:59,800
not saying large language models
have 0 use cases. 

564
00:24:59,800 --> 00:25:01,640
That's a mistake I've made a few
years ago. 

565
00:25:01,640 --> 00:25:03,680
I really did. 
There are specific niche. 

566
00:25:03,760 --> 00:25:05,400
It's cloud software, it's cloud 
compute like. 

567
00:25:05,600 --> 00:25:07,840
There are things that can do. 
There are some interesting 

568
00:25:07,840 --> 00:25:10,400
things people can do with code. 
Can it write in type? 

569
00:25:10,400 --> 00:25:13,280
No, you can't rely on it. 
The future of that will be in 

570
00:25:13,280 --> 00:25:16,320
what is the DG X boxes that 
NVIDIA they're making like the 

571
00:25:17,040 --> 00:25:19,360
like, not home, but like 
researcher grade extremely 

572
00:25:19,360 --> 00:25:21,600
expensive computers. 
There is something there. 

573
00:25:21,800 --> 00:25:24,560
The thing is, man, I don't know 
what quadrant and also Casey 

574
00:25:24,560 --> 00:25:26,960
Newton needs to go and read that
Kilo Code blog about how 

575
00:25:26,960 --> 00:25:29,560
inference costs are going up 
because he said the opposite 

576
00:25:29,560 --> 00:25:32,480
would happen. 
Who any Sorry, sorry, Casey 

577
00:25:32,480 --> 00:25:33,920
Newton said. 
That the cost of inference. 

578
00:25:34,000 --> 00:25:36,400
There's a blog from Kilo Code 
that shows how the cost of 

579
00:25:36,400 --> 00:25:38,520
inference is going up, because 
while the prices might be 

580
00:25:38,520 --> 00:25:40,840
reducing, the amount of 
inference the coding models are 

581
00:25:40,840 --> 00:25:43,200
using is actually massive. 
Yeah, because because of deep, 

582
00:25:43,280 --> 00:25:45,640
because of learning, because, 
because of reasoning models and 

583
00:25:45,640 --> 00:25:47,880
and that suddenly kind of turned
inferencing into because. 

584
00:25:48,000 --> 00:25:49,520
There's more. 
Training like experience. 

585
00:25:49,680 --> 00:25:53,000
They're trying to do more stuff,
yes, but like the old models, 

586
00:25:53,000 --> 00:25:55,480
doing the same thing certainly 
costs less. 

587
00:25:55,480 --> 00:25:57,720
Except no one wants to use the 
old models. 

588
00:25:57,760 --> 00:26:00,160
No one. 
Wants to advancing so much that 

589
00:26:00,160 --> 00:26:01,920
people are excited about the new
model. 

590
00:26:01,920 --> 00:26:03,440
Well, let's actually talk about 
that for a second. 

591
00:26:03,440 --> 00:26:05,920
It's like a damned if you are 
do, damned if you don't 

592
00:26:05,920 --> 00:26:07,240
situation like. 
Every every time. 

593
00:26:07,240 --> 00:26:10,200
Why is it that I have the 
negative No, and I mean there's 

594
00:26:10,200 --> 00:26:12,240
not an attack on you just. 
Because you're you're, you're 

595
00:26:12,520 --> 00:26:17,040
opining on AI like when you're. 
But I don't feel like this level

596
00:26:17,040 --> 00:26:19,600
of rigor, and I'm not talking 
about you here. 

597
00:26:19,840 --> 00:26:23,040
I don't feel like this level of 
rigor and criticism and analysis

598
00:26:23,040 --> 00:26:26,440
is ever put on the positive side
to make them prove themselves. 

599
00:26:26,600 --> 00:26:29,000
Indeed, I don't. 
You can say damned if I'm. 

600
00:26:29,600 --> 00:26:31,480
They've never had to sing for 
their supper. 

601
00:26:31,640 --> 00:26:34,440
They've never had to really sit 
there and explain what their 

602
00:26:34,440 --> 00:26:36,920
product does. 
You yourself even said it's for 

603
00:26:36,920 --> 00:26:40,560
users to work out why. 
Like they're worth 300 billion. 

604
00:26:40,560 --> 00:26:44,520
That's more than the market cap 
of Salesforce, $500 billion more

605
00:26:44,520 --> 00:26:47,200
than the market cap or around 
the market cap of Netflix. 

606
00:26:47,400 --> 00:26:49,880
Yeah, they do actually kind of 
have to justify themselves. 

607
00:26:50,040 --> 00:26:53,440
If they are worth that much, 
then yeah, I actually would like

608
00:26:53,440 --> 00:26:55,040
an explanation. 
Well, but you're talking about 

609
00:26:55,040 --> 00:26:57,360
different justifications here 
because you know, there's 

610
00:26:57,360 --> 00:27:00,000
justification to investors and 
then justification to clients, 

611
00:27:00,000 --> 00:27:03,200
to customers about whether this 
this software has any utility. 

612
00:27:03,200 --> 00:27:06,000
And and I think you are seeing a
huge amount of resistance from 

613
00:27:06,120 --> 00:27:08,440
larger enterprises at least over
whether they really want to 

614
00:27:08,440 --> 00:27:10,600
adopt this stuff. 
Like, you know, Steve, Laura had

615
00:27:10,600 --> 00:27:14,480
a pretty interesting piece that 
came out today that basically 

616
00:27:14,480 --> 00:27:17,280
said, I mean, you know, and this
was not really a case fully 

617
00:27:17,280 --> 00:27:19,520
against AI, but basically said 
we're in like the through of 

618
00:27:19,520 --> 00:27:22,280
disillusionment right now when 
it comes to the number of 

619
00:27:22,360 --> 00:27:26,000
products that are being, you 
know, abandoned by enterprises 

620
00:27:26,000 --> 00:27:28,240
right now, you know, initiatives
that they end up abandoning. 

621
00:27:28,240 --> 00:27:31,520
It's like 42% or something. 
So I actually don't really think

622
00:27:31,520 --> 00:27:35,360
these guys are not out there not
not being called to by the CIOs 

623
00:27:35,360 --> 00:27:37,720
of all these prospective 
customers to say like what is 

624
00:27:37,720 --> 00:27:40,880
the value this has here? 
Even Tyler Cowen had a piece 

625
00:27:40,880 --> 00:27:44,400
about this that was ultimately 
optimistic on AI, but that the 

626
00:27:44,400 --> 00:27:47,840
utility point of where these 
corporate customers and buyers 

627
00:27:47,840 --> 00:27:50,840
hadn't, you know, fully adopted 
it yet was simply just because 

628
00:27:50,840 --> 00:27:53,040
it was too early. 
It's not like this 

629
00:27:53,120 --> 00:27:54,760
transformation should take a 
long time. 

630
00:27:56,160 --> 00:27:59,880
It's only been three years. 
AI yeah, it's been very fat. 

631
00:28:00,400 --> 00:28:03,040
You do like cloud. 
There are banks that still don't

632
00:28:03,040 --> 00:28:04,720
embrace the cloud like there are
plenty. 

633
00:28:04,720 --> 00:28:07,960
Of 2017 was attention all you 
need it's actually been much 

634
00:28:07,960 --> 00:28:09,000
longer and on. 
Top of that. 

635
00:28:09,360 --> 00:28:13,560
You've never had this focused 
capital. 2022 and like never 

636
00:28:13,560 --> 00:28:16,760
just catching on to in 2023. 
You've never had this focused 

637
00:28:16,760 --> 00:28:20,800
capital in investment history. 
You have never had all the 

638
00:28:20,800 --> 00:28:22,600
King's horses and all the King's
men do this. 

639
00:28:22,760 --> 00:28:26,320
It's you have most of the talent
and most of the money diverted 

640
00:28:26,320 --> 00:28:28,960
to focus on one thing. 
You've got start-ups, you've got

641
00:28:29,080 --> 00:28:32,160
giant companies, you've got 
everyone looking at this and 

642
00:28:32,160 --> 00:28:34,720
this is what we've got. 
What is diverted? 

643
00:28:34,720 --> 00:28:36,920
I think this is actually a key 
point of your argument that I 

644
00:28:36,920 --> 00:28:37,880
understand. 
Like what? 

645
00:28:38,080 --> 00:28:40,480
What do you want the money to go
to instead? 

646
00:28:40,560 --> 00:28:42,120
I guess like the way. 
Where else? 

647
00:28:42,120 --> 00:28:45,320
Capitalism works is like this is
the most exciting idea people 

648
00:28:45,320 --> 00:28:47,080
have going. 
Like if they had another great 

649
00:28:47,080 --> 00:28:49,200
idea, they could return a bunch 
of money. 

650
00:28:49,240 --> 00:28:50,800
They they are happy to hear you 
out. 

651
00:28:50,800 --> 00:28:53,040
Like what is the better? 
What is the thing? 

652
00:28:53,480 --> 00:28:55,400
What is the thing you want them 
to spend the money on? 

653
00:28:55,480 --> 00:28:56,720
I mean. 
This. 

654
00:28:56,920 --> 00:28:59,640
Anything else? 
I mean, genuinely, how about a 

655
00:28:59,640 --> 00:29:02,200
product that makes more money 
than it costs? 

656
00:29:02,320 --> 00:29:05,120
How about, I don't know, 
something that helps niche 

657
00:29:05,120 --> 00:29:07,040
industries? 
That's the thing, I'm not. 

658
00:29:08,240 --> 00:29:10,640
Like the the biology AI 
movement. 

659
00:29:10,640 --> 00:29:12,120
You know, like AI for drugs. 
Which one? 

660
00:29:12,360 --> 00:29:14,280
Which I guess in some ways. 
I think that's cool. 

661
00:29:14,520 --> 00:29:16,480
I think there are tons of AI 
things that. 

662
00:29:16,560 --> 00:29:19,840
Build in this space, yeah. 
I think that those are not all 

663
00:29:19,840 --> 00:29:22,640
large language model products, 
and indeed, if large language 

664
00:29:22,640 --> 00:29:25,000
models make money and do that, 
that's sick. 

665
00:29:25,080 --> 00:29:28,200
That would be cool. 
I'm not saying that I can't find

666
00:29:28,200 --> 00:29:31,800
a thing I like, I'm just yet to 
be remotely impressed by this. 

667
00:29:32,000 --> 00:29:36,480
And again, it comes from the 
fact that every really, I really

668
00:29:36,480 --> 00:29:39,120
encourage you to go and look, 
look at the early coverage of 

669
00:29:39,120 --> 00:29:42,240
this and then follow the 
coverage year by year and see 

670
00:29:42,320 --> 00:29:46,560
how vague this has been and how 
ridiculous the pressure has been

671
00:29:46,560 --> 00:29:48,680
on people. 
Look outside of the tech media 

672
00:29:48,680 --> 00:29:51,080
into like the Business Media, 
into industrial media, 

673
00:29:51,200 --> 00:29:55,560
educational media, put the 
pressure on teachers from 20/22.

674
00:29:55,720 --> 00:29:58,960
It is absolutely ridiculous the 
amount of pressure put on 

675
00:29:58,960 --> 00:30:02,240
society, the amount of pressure 
from bosses telling people 

676
00:30:02,400 --> 00:30:04,760
you've got to do AI. 
It's the same crap they were 

677
00:30:04,760 --> 00:30:08,120
doing with return to office. 
It is the same thing. 

678
00:30:08,280 --> 00:30:10,400
And you and you same. 
And you hold the media 

679
00:30:10,400 --> 00:30:11,920
responsible for that for a lot 
of that. 

680
00:30:11,920 --> 00:30:15,200
Partially, I think that there is
also a societal failure in 

681
00:30:15,200 --> 00:30:17,320
general. 
I think that the markets are not

682
00:30:17,320 --> 00:30:21,120
connected to real outcomes and I
think that that rewards a 

683
00:30:21,120 --> 00:30:23,640
certain kind of burn. 
I also think that there is a 

684
00:30:23,880 --> 00:30:27,040
there is a level here of non 
like it's not malicious 

685
00:30:27,040 --> 00:30:28,720
necessarily. 
It's like venture capital did 

686
00:30:28,720 --> 00:30:31,080
what venture capital would do. 
Big tech did what big tech do, 

687
00:30:31,160 --> 00:30:33,760
open AI did what any startup 
would do. 

688
00:30:33,880 --> 00:30:35,960
I don't think they set out being
like we, we are going to do a 

689
00:30:35,960 --> 00:30:38,080
big fake thing and we will get 
everyone to. 

690
00:30:38,080 --> 00:30:39,520
But I don't think that's it at 
all. 

691
00:30:39,520 --> 00:30:42,040
I think that this is very much 
something that ran out. 

692
00:30:42,200 --> 00:30:45,680
It kind of got out of control 
versus a malignant thing where 

693
00:30:45,680 --> 00:30:49,200
Sam Alton was sitting, you know,
Finally, no, I think it's just a

694
00:30:49,200 --> 00:30:51,080
runaway train. 
Yeah, no. 

695
00:30:51,280 --> 00:30:54,480
And I think an argument in that 
favour is the fact that ChatGPT 

696
00:30:54,480 --> 00:30:56,920
as a product was kind of a 
mistaken success. 

697
00:30:56,920 --> 00:30:59,960
You know they put it out on a 
lark, pissing off everyone at 

698
00:30:59,960 --> 00:31:01,760
Microsoft. 
Because they wanted to see. 

699
00:31:02,360 --> 00:31:04,800
Yeah, but yeah. 
Isn't that like an argument for 

700
00:31:04,800 --> 00:31:06,800
this not being some marketing 
coup? 

701
00:31:06,800 --> 00:31:08,640
It was like. 
But it is saying is that it is. 

702
00:31:08,640 --> 00:31:11,000
What they had and people were 
excited about it. 

703
00:31:11,000 --> 00:31:13,160
No, but but I think what he's 
saying is it it got out of 

704
00:31:13,160 --> 00:31:16,160
control because it grew out of 
this pace that is ultimately 

705
00:31:16,160 --> 00:31:18,160
unsustainable or based largely 
on hype. 

706
00:31:18,160 --> 00:31:20,920
And so you know, the products 
utility doesn't really need a 

707
00:31:20,920 --> 00:31:23,680
type, even though so there were.
Other companies that did what 

708
00:31:23,680 --> 00:31:26,720
ChatGPT did before them, that 
was like, I think Jasper beat 

709
00:31:26,720 --> 00:31:28,960
them to market. 
They didn't have they didn't 

710
00:31:28,960 --> 00:31:31,160
have Sam Altman's cachet. 
And I mean that. 

711
00:31:31,160 --> 00:31:34,000
And that really is just 
start-ups that I'm not going to 

712
00:31:34,000 --> 00:31:35,560
be pissy about. 
The product market thing can 

713
00:31:35,560 --> 00:31:37,080
happen in lots of different 
ways, I mean. 

714
00:31:37,080 --> 00:31:38,480
Yeah, and it happens all the 
time. 

715
00:31:38,480 --> 00:31:41,480
It's just like, yeah, it was a 
marketing thing and it was 

716
00:31:41,480 --> 00:31:44,200
something that people got 
legitimately excited about at 

717
00:31:44,200 --> 00:31:45,560
first. 
I'm not saying any of that was 

718
00:31:45,560 --> 00:31:47,360
illegitimate. 
I actually think that the 

719
00:31:47,360 --> 00:31:49,720
original thing that it did was 
kind of interesting. 

720
00:31:49,720 --> 00:31:51,760
It's like, good. 
I remember at the time it didn't

721
00:31:51,760 --> 00:31:54,920
really get a ton more 
interesting and indeed the 

722
00:31:54,920 --> 00:31:57,840
things it can do did not 
dramatically change. 

723
00:31:58,120 --> 00:32:01,320
And I even think that the launch
of GPT 5 when you look at the 

724
00:32:01,560 --> 00:32:05,600
craziness around 4 O it reminds 
me and someone on blue sky. 

725
00:32:05,640 --> 00:32:07,920
At this point it's like it's 
like a live service game. 

726
00:32:08,120 --> 00:32:09,920
People freaked out like gamers 
do. 

727
00:32:09,920 --> 00:32:12,560
Being like you took away the 
thing and now you've got people 

728
00:32:12,560 --> 00:32:16,360
on Reddit saying 4 O isn't the 
same, which is just it's full 

729
00:32:16,360 --> 00:32:18,960
but it suggests that there is 
not a product people are 

730
00:32:18,960 --> 00:32:23,080
attached to but a feeling the 
thing it makes the it's almost 

731
00:32:23,080 --> 00:32:25,440
like it. 
Isn't that like, OK, again, I 

732
00:32:25,440 --> 00:32:29,720
think this fits in this sort of.
It's either very successful at 

733
00:32:29,720 --> 00:32:32,600
being bad or bad at being 
successful and you can sort of 

734
00:32:32,600 --> 00:32:35,520
switch back and forth. 
I'm not saying it's not popular,

735
00:32:35,680 --> 00:32:39,560
I'm just saying that the it's 
popularity is kind of a mirage. 

736
00:32:39,560 --> 00:32:42,480
First of all, the numbers that 
Open AI are giving us are 

737
00:32:42,680 --> 00:32:45,520
questionable, and I've 
repeatedly asked them to give 

738
00:32:45,520 --> 00:32:49,280
clarity over what a weekly user 
is, whether they have monthly 

739
00:32:49,280 --> 00:32:51,400
active users. 
They have revealed their annual 

740
00:32:51,400 --> 00:32:55,200
recurring revenue once to CNBC 
on June 3rd I believe, and then 

741
00:32:55,200 --> 00:32:58,520
leaked twice within 48 hours to 
two different outlets when they 

742
00:32:58,520 --> 00:33:01,040
hit 12 and 13. 
That alone suggests that the way

743
00:33:01,040 --> 00:33:02,840
they calculate annualized 
revenue is weird. 

744
00:33:03,080 --> 00:33:06,400
The thing is, I do think that 
media pressure and I do think 

745
00:33:06,400 --> 00:33:09,040
societal pressure can lead to 
this thing and the fact that 

746
00:33:09,040 --> 00:33:12,280
there are no other companies 
that really show any kind of 

747
00:33:12,280 --> 00:33:16,160
similar revenue scale, Even big 
tech really isn't. 

748
00:33:16,320 --> 00:33:20,160
So this can be popular. 
This can be used by a lot of 

749
00:33:20,160 --> 00:33:23,840
people, but there are no after 
effects that suggest that the 

750
00:33:23,840 --> 00:33:27,640
large language model is popular.
The thing, the headless beast 

751
00:33:27,640 --> 00:33:31,040
ChatGPT is popular, but why is 
it popular? 

752
00:33:31,040 --> 00:33:33,200
What do people, they use it as 
search, They use it to 

753
00:33:33,200 --> 00:33:36,280
brainstorm and whatever. 
I think that the therapy thing 

754
00:33:36,280 --> 00:33:38,120
is very worrying, as is the 
friendship thing, but 

755
00:33:38,120 --> 00:33:40,040
nevertheless, it's the thing 
that people use it for. 

756
00:33:40,160 --> 00:33:43,000
But when it comes down to the 
actual utility, when it comes 

757
00:33:43,000 --> 00:33:45,360
down to the thing that makes 
this worth this much money, 

758
00:33:45,480 --> 00:33:47,760
that's where my that's where the
wheels come off for me. 

759
00:33:47,880 --> 00:33:50,880
Even the coding stuff, right. 
That's what I thought that that 

760
00:33:50,880 --> 00:33:54,800
might be a huge revenue thing, 
but 400 million ARR from Claude 

761
00:33:54,800 --> 00:33:57,320
code according to the 
information, that's stinky. 

762
00:33:57,440 --> 00:34:00,000
That's stinky, stinky. 
I mean, that's over four months 

763
00:34:00,000 --> 00:34:01,040
though. 
I mean like they just. 

764
00:34:01,040 --> 00:34:01,720
Lost. 
Yeah, and it's gonna. 

765
00:34:01,840 --> 00:34:04,120
And they're now rate limiting 
heavily. 

766
00:34:04,120 --> 00:34:06,720
These are some of the fastest 
growing products by revenue 

767
00:34:06,960 --> 00:34:09,360
like. 
And then they will crash because

768
00:34:09,360 --> 00:34:11,400
cursor you want to talk. 
Cursor I'd love to talk. 

769
00:34:11,400 --> 00:34:12,719
Cursor, I'd love to talk. 
Cursor. 

770
00:34:12,719 --> 00:34:14,840
We can talk cursor. 
I'd love to talk Cursor. 

771
00:34:14,840 --> 00:34:17,800
Anyway, I think Cursor is the 
best earns of this entire 

772
00:34:17,800 --> 00:34:20,239
situation they released. 
Do you know anything that 

773
00:34:20,239 --> 00:34:22,080
happened with cursors? 
Price change? 

774
00:34:22,159 --> 00:34:23,480
Yeah. 
I wrote a whole piece about it 

775
00:34:23,639 --> 00:34:25,000
on Sunday. 
Wonderful, I should read that 

776
00:34:25,000 --> 00:34:25,800
shit. 
I'm so sorry. 

777
00:34:25,920 --> 00:34:28,239
It's okay, it's fine. 
Anyway, have you noticed? 

778
00:34:28,239 --> 00:34:30,560
The afterwards. 
Are you familiar with the auto 

779
00:34:30,719 --> 00:34:32,679
feature on cursor? 
Yes, yes, sure. 

780
00:34:32,679 --> 00:34:34,880
Not how. 
Unlimited, Yes, not unlimited. 

781
00:34:34,880 --> 00:34:37,600
It's starting in September. 
They're taking that cursor has 

782
00:34:37,600 --> 00:34:39,159
had to change their entire 
product. 

783
00:34:39,480 --> 00:34:41,840
They've had to change everything
and now even the unlimited part 

784
00:34:41,840 --> 00:34:45,040
isn't unlimited. 
That 400, sorry, 500 million 

785
00:34:45,040 --> 00:34:47,880
annualized figure was from May, 
I want to say. 

786
00:34:47,920 --> 00:34:49,679
It's from June, they're above 
that now. 

787
00:34:49,920 --> 00:34:50,880
Aren't they? 
Yeah. 

788
00:34:50,880 --> 00:34:52,120
How long are they going to tell 
you about that? 

789
00:34:52,120 --> 00:34:54,560
What's the churn? 
Because The thing is that 

790
00:34:54,560 --> 00:34:56,440
product has dramatically 
changed. 

791
00:34:56,520 --> 00:34:59,560
It's no longer the same thing. 
I read their forums and they 

792
00:34:59,560 --> 00:35:02,280
read it all the time. 
They are in September 

793
00:35:02,480 --> 00:35:05,080
dramatically changing again. 
They're basically making it so 

794
00:35:05,080 --> 00:35:07,560
you will be hard stopped on 
cursor like you are with Claude 

795
00:35:07,560 --> 00:35:08,520
code. 
Yeah. 

796
00:35:08,520 --> 00:35:10,400
Well, their margins are gross 
margins were negative. 

797
00:35:10,400 --> 00:35:11,680
I mean it was a nightmare for 
them. 

798
00:35:12,160 --> 00:35:15,120
They had these sinkhole 
customers that were using all 

799
00:35:15,120 --> 00:35:16,960
these agentic tools that were 
burning all these. 

800
00:35:17,680 --> 00:35:20,880
They basically were propping up 
Anthropics business more than 

801
00:35:20,880 --> 00:35:23,840
10% at times of Anthropics 
revenue comes from is that. 

802
00:35:24,080 --> 00:35:25,840
True. 
Yeah, Yeah. 

803
00:35:26,680 --> 00:35:29,720
All a newcomer.co, sure. 
I need to save him another 

804
00:35:29,720 --> 00:35:30,680
negative point. 
He's like. 

805
00:35:30,680 --> 00:35:32,520
I'll add that to my quiver of 
negative. 

806
00:35:33,760 --> 00:35:35,440
I will, Eric. 
The important thing here is that

807
00:35:35,440 --> 00:35:37,240
my work gets more widely 
distributed. 

808
00:35:37,240 --> 00:35:38,640
That's what, that's my 
contribution. 

809
00:35:39,120 --> 00:35:41,720
That's. 
The questions about cursor too, 

810
00:35:41,720 --> 00:35:44,040
we report it. 
I mean, I'm not a super, I mean,

811
00:35:44,040 --> 00:35:46,520
I'm fairly bullish about the, 
you know, what's happening with 

812
00:35:46,520 --> 00:35:49,680
language models overall, but 
obviously we believe in asking 

813
00:35:49,680 --> 00:35:51,560
questions and quibbling with 
things. 

814
00:35:51,560 --> 00:35:52,760
So we're supportive with you on 
that. 

815
00:35:52,760 --> 00:35:57,080
I still think the bulls, I wish,
I wish 20% of what they said was

816
00:35:57,080 --> 00:36:00,520
negative and you sort of the 
bear, I wish, you know, there 

817
00:36:00,520 --> 00:36:03,040
was 20% that you would offer 
that's positive. 

818
00:36:03,080 --> 00:36:05,720
And I'm not clear what you're 
arguing for. 

819
00:36:05,720 --> 00:36:08,640
Sometimes, I mean. 
I'm arguing for us to stop 

820
00:36:08,640 --> 00:36:10,440
pretending that it's something 
it isn't. 

821
00:36:10,640 --> 00:36:14,160
I'm arguing for people to stop 
saying that the agents work. 

822
00:36:14,160 --> 00:36:17,400
I'm arguing for people to stop 
pretending the open AI is a 

823
00:36:17,400 --> 00:36:19,560
sustainable company. 
Same deal with Anthropic. 

824
00:36:19,880 --> 00:36:22,440
For people to admit the 
economics are rotten and also 

825
00:36:22,560 --> 00:36:25,040
start talking about the fact 
that the money isn't big, not 

826
00:36:25,040 --> 00:36:26,520
just. 
But like I mean you were you 

827
00:36:26,520 --> 00:36:30,320
were negative on like Facebook 
and I think like 2022, which was

828
00:36:30,480 --> 00:36:31,960
like the low point for the 
stock. 

829
00:36:31,960 --> 00:36:35,280
The stock is like 7X from there 
and like the map public. 

830
00:36:35,280 --> 00:36:39,120
Stocks are different to product.
Producing profits, they make 

831
00:36:39,200 --> 00:36:41,480
their generate products, things 
like a. 

832
00:36:41,480 --> 00:36:44,040
Big asshole. 
I'm so sorry. 

833
00:36:44,040 --> 00:36:47,200
Are we talking about? 
Are we talking about companies 

834
00:36:47,360 --> 00:36:48,840
do you believe in? 
Capitalism. 

835
00:36:49,000 --> 00:36:52,880
Yes, Oh my God, it exists. 
Just a sincere question, not to 

836
00:36:52,880 --> 00:36:54,440
like. 
I believe that the current state

837
00:36:54,440 --> 00:36:56,920
of capitalism with with rot 
economics, the growth of all 

838
00:36:56,920 --> 00:36:59,880
costs ecosystem is going to 
drive the tech industry into the

839
00:36:59,880 --> 00:37:02,320
goddamn toiler. 
I think it is going to end up 

840
00:37:02,320 --> 00:37:05,400
hurting far more people and it 
goes all the way back to 2011 

841
00:37:05,400 --> 00:37:08,160
with the case of the fat startup
software is eating the world was

842
00:37:08,160 --> 00:37:11,720
not a good thing for startups. 
All it did was attempt to remove

843
00:37:11,720 --> 00:37:14,600
the connection from value from 
the value of the stock. 

844
00:37:14,800 --> 00:37:17,680
That was the purpose of software
is eating the world. 

845
00:37:17,920 --> 00:37:20,760
That is my problem. 
The the Bay Area that was about 

846
00:37:20,760 --> 00:37:24,160
creating valuable software that 
you sell to people and then you 

847
00:37:24,160 --> 00:37:26,880
sell to a big company for a 
profit or you take public is 

848
00:37:27,080 --> 00:37:30,000
dead. 
It is absolutely dead as one of.

849
00:37:30,000 --> 00:37:34,040
The in the AI world right now is
the cool calling card is to do 

850
00:37:34,040 --> 00:37:36,200
something with as few employees 
as possible. 

851
00:37:36,200 --> 00:37:41,000
Like there there was this period
of bloat, but I do think a lot 

852
00:37:41,120 --> 00:37:42,920
people are trying to like skip 
funny rounds. 

853
00:37:42,920 --> 00:37:44,560
People are trying to hire fewer 
people. 

854
00:37:44,560 --> 00:37:47,840
Like I do think some of the 
values of the current AI boom 

855
00:37:47,840 --> 00:37:51,600
are in sort of lean companies, 
obviously. 

856
00:37:51,640 --> 00:37:54,280
And do you think the economics 
will always be? 

857
00:37:54,600 --> 00:37:57,520
There for that because I hope so
I hope that someone finds a 

858
00:37:57,520 --> 00:37:59,400
thing to do with large language 
models that's actually 

859
00:37:59,400 --> 00:38:01,360
profitable while not destroying 
the environment. 

860
00:38:01,640 --> 00:38:05,240
I really there are interesting 
things Simon Willis and Max Wolf

861
00:38:05,240 --> 00:38:07,080
they post cool things all the 
time. 

862
00:38:07,080 --> 00:38:09,760
I can't do them, can't code, but
like, I recognize that there are

863
00:38:09,760 --> 00:38:13,120
people doing stuff for that, but
there's this massive other story

864
00:38:13,360 --> 00:38:15,840
I don't ever want to read about 
fucking agents the rest of my 

865
00:38:16,640 --> 00:38:19,720
phase until they can. 
Until like the ideally early 

866
00:38:19,760 --> 00:38:22,280
yes, it still doesn't value like
through these things while 

867
00:38:22,280 --> 00:38:24,120
they're. 
Look at the coverage of Open AI.

868
00:38:24,200 --> 00:38:27,360
I mean, is deep research an 
agent like deep research? 

869
00:38:27,440 --> 00:38:28,920
Man, no it's not. 
Good. 

870
00:38:28,920 --> 00:38:30,960
Like a consulting memo. 
I'm sure you're like, well 

871
00:38:30,960 --> 00:38:33,400
consultants are idiots but like 
if you want to deliver like a 

872
00:38:33,400 --> 00:38:36,520
consulting memo level product, 
you could get it out of deep 

873
00:38:36,520 --> 00:38:39,480
research. 
Can you like cat? 

874
00:38:39,480 --> 00:38:41,720
Like is that something like 
here's the thing. 

875
00:38:41,720 --> 00:38:44,440
I've read a few articles about 
McKinsey being scared of AI. 

876
00:38:44,720 --> 00:38:47,720
There's never really any numbers
or tangible. 

877
00:38:48,160 --> 00:38:50,480
There's always like, well, this 
is obviously happening. 

878
00:38:50,480 --> 00:38:53,200
But when you go and look for 
tangible examples, in fact, 

879
00:38:53,200 --> 00:38:55,840
throughout this whole boom, 
they're just not there. 

880
00:38:56,040 --> 00:38:57,760
And that's my but that's the 
thing. 

881
00:38:57,920 --> 00:39:00,560
If this was being fairly 
evaluated, great. 

882
00:39:00,840 --> 00:39:04,400
It isn't. 
And I'm roasted like a prize hog

883
00:39:04,440 --> 00:39:08,040
for daring to step outside here,
for daring to say this. 

884
00:39:08,160 --> 00:39:11,480
But they did not receive this 
level of scrutiny on the way up.

885
00:39:11,680 --> 00:39:14,240
They have not really received 
this level of scrutiny. 

886
00:39:14,520 --> 00:39:17,920
And if I'm right, there was a 
bloody reason to and that this 

887
00:39:17,920 --> 00:39:21,800
is kind of a giant mirage. 
And it's scary because putting 

888
00:39:21,800 --> 00:39:26,000
aside my personal feelings, the 
the economic effects here and to

889
00:39:26,000 --> 00:39:28,160
the effects for startups, 
because the thing I said 

890
00:39:28,160 --> 00:39:31,120
earlier, I've set this in the AI
as a money trap. 

891
00:39:31,120 --> 00:39:33,680
Putting aside feelings about the
industry, we currently have a 

892
00:39:33,680 --> 00:39:36,760
crisis with valuations. 
Who is going to buy Cognition? 

893
00:39:36,760 --> 00:39:39,240
Who is going to buy Cohere? 
Who is going to buy Cursor? 

894
00:39:39,440 --> 00:39:41,280
Is Cursor going to sell for $20 
billion? 

895
00:39:41,360 --> 00:39:43,280
OK, fine. 
Will they actually sell for the 

896
00:39:43,280 --> 00:39:45,440
IP though? 
Who is going to buy these 

897
00:39:45,440 --> 00:39:47,640
things? 
Why did Cognition by the remains

898
00:39:47,640 --> 00:39:51,280
of Windsurf? 
Was it to add that 782,000,000 

899
00:39:51,440 --> 00:39:53,560
ARR? 
Because condition does not have 

900
00:39:53,800 --> 00:39:55,840
revenue. 
What are these companies doing? 

901
00:39:55,840 --> 00:39:58,600
What happens next? 
How do they exit? 

902
00:39:58,600 --> 00:40:01,800
Because remember in 2021 a bunch
of crap money got tied up in 

903
00:40:01,800 --> 00:40:04,640
crapper companies and LP's got 
pissy. 

904
00:40:04,800 --> 00:40:07,120
LP's came flooding back in with 
AI. 

905
00:40:07,440 --> 00:40:09,720
How do you think they're going 
to feel when there's no exit? 

906
00:40:10,120 --> 00:40:11,760
Because there really isn't right
now. 

907
00:40:11,760 --> 00:40:14,080
Inflection wasn't a real 1 
character. 

908
00:40:14,080 --> 00:40:17,440
That AI wasn't the shareholders.
The investors are getting cash 

909
00:40:17,440 --> 00:40:19,240
back because of those. 
Which investors? 

910
00:40:19,240 --> 00:40:20,680
So they are some. 
Some are, some are. 

911
00:40:20,680 --> 00:40:24,520
Investors and companies like 
Inflection Character, they're 

912
00:40:24,520 --> 00:40:26,120
getting money back as part of 
those deals. 

913
00:40:26,120 --> 00:40:28,640
But they they are not getting 
the massive returns that would 

914
00:40:28,640 --> 00:40:30,200
justify like. 
That's the thing. 

915
00:40:30,360 --> 00:40:32,320
It's. 
Some of them are like 2X and 

916
00:40:32,320 --> 00:40:34,800
pretty big amounts of money. 
I mean it depends on the. 

917
00:40:34,800 --> 00:40:39,080
DOX is that really like from the
again, if this was just a 

918
00:40:39,080 --> 00:40:42,480
regular thing, If this is a 
regular thing where I was not 

919
00:40:42,480 --> 00:40:45,680
having conversations with people
that get animated about it. 

920
00:40:45,800 --> 00:40:48,960
If it was just a regular stuff 
something, hell yeah, 2X3X cool.

921
00:40:49,160 --> 00:40:51,720
But the weird thing is, is 
everyone's talking about this is

922
00:40:51,720 --> 00:40:54,280
the next big thing. 
When you look at the look at the

923
00:40:54,280 --> 00:40:56,920
actual things. 
Why didn't these companies get 

924
00:40:56,920 --> 00:41:00,440
acquired properly? 
Why did they do these acqui hire

925
00:41:00,440 --> 00:41:02,440
thingies? 
That's the financial. 

926
00:41:02,760 --> 00:41:04,640
There's a whole, there's a 
regulatory whole angle. 

927
00:41:04,640 --> 00:41:06,720
I will say there is a whole. 
Regular there's a regulatory. 

928
00:41:06,720 --> 00:41:09,320
Angle that happen. 
Won't regulate like you think. 

929
00:41:09,320 --> 00:41:11,880
They're actually Microsoft's 
inflection deal got put through 

930
00:41:11,880 --> 00:41:13,560
regulators anyway, like and it 
passed. 

931
00:41:13,560 --> 00:41:14,960
No, I mean, it did, yeah. 
It's. 

932
00:41:15,160 --> 00:41:18,320
Used to avoid regulation. 
It it did after the fact, but 

933
00:41:18,320 --> 00:41:20,160
the deal sailed through. 
I mean, you know, it was just 

934
00:41:20,160 --> 00:41:22,080
basically a licensing deal. 
I mean, that's the reason these 

935
00:41:22,080 --> 00:41:24,200
things caught fire is because. 
We did the deal. 

936
00:41:24,280 --> 00:41:26,800
These things are legal. 
Tricks to get around regulation.

937
00:41:27,080 --> 00:41:28,840
Yeah, these aren't the exits 
people want. 

938
00:41:29,080 --> 00:41:32,280
But I'm saying right, people are
making money off them and an 

939
00:41:32,280 --> 00:41:34,760
open AI is selling like 
secondary shares for employees 

940
00:41:34,760 --> 00:41:37,560
at like 500, like an insane 
number. 

941
00:41:37,560 --> 00:41:41,080
Like yeah, it's like hard to 
even say. 

942
00:41:41,120 --> 00:41:44,000
I'm like, Oh my God, isn't like.
Worth this much or so? 

943
00:41:44,040 --> 00:41:46,120
But like I think I. 
Think to make as an argument 

944
00:41:46,120 --> 00:41:48,280
here, I think what's going on 
here with these like kind of, 

945
00:41:48,520 --> 00:41:51,320
you know, mild outcomes for the 
second tier startups is that 

946
00:41:51,560 --> 00:41:53,480
they're all kind of predicate on
the belief that there is going 

947
00:41:53,480 --> 00:41:55,520
to be 1, maybe 2 actual huge 
outcomes. 

948
00:41:55,520 --> 00:41:57,400
And it's probably open AI and 
it's anthropic. 

949
00:41:57,640 --> 00:42:00,120
And if those do, if and if they 
don't turn out to be anything, 

950
00:42:00,120 --> 00:42:01,520
then yeah, the whole thing does.
Probably. 

951
00:42:01,640 --> 00:42:03,880
I mean, that was a big thing in 
my benchmark piece right there 

952
00:42:03,880 --> 00:42:05,480
that I just published. 
You know, it's like, oh, they're

953
00:42:05,480 --> 00:42:07,440
making a bunch of infrastructure
investments, but they don't have

954
00:42:07,440 --> 00:42:09,880
a big foundation model. 
And it's very possible there's a

955
00:42:09,880 --> 00:42:12,360
classic home run distribution. 
But to me that argument is 

956
00:42:12,360 --> 00:42:14,920
paired with open AI is going to,
you know, win everything in the 

957
00:42:14,920 --> 00:42:16,800
Super value. 
Actually this is a really good 

958
00:42:16,800 --> 00:42:19,080
question though. 
What happens if open AI doesn't 

959
00:42:19,080 --> 00:42:21,040
convert to a non profit and 
doesn't receive the money from 

960
00:42:21,040 --> 00:42:22,920
exactly? 
Well, I agree we are aligned on 

961
00:42:22,920 --> 00:42:23,720
that. 
You know I wrote a. 

962
00:42:23,720 --> 00:42:25,960
Bare thing. 
About open AI predicated on on 

963
00:42:25,960 --> 00:42:27,240
that. 
So I agree with you on that. 

964
00:42:27,240 --> 00:42:31,440
Banger And The thing is, I don't
actually think we disagree on a 

965
00:42:31,440 --> 00:42:34,720
ton of, I think we have, you 
have a lot of hopes for this, 

966
00:42:34,720 --> 00:42:36,920
but it seems like the actual 
economics, you kind of see where

967
00:42:36,920 --> 00:42:39,200
I'm coming from. 
And I think I think that it's 

968
00:42:39,200 --> 00:42:41,920
good that Someone Like You, it 
has the bet, has the bull 

969
00:42:41,920 --> 00:42:45,120
argument, but still acknowledges
these things because whoever's 

970
00:42:45,120 --> 00:42:47,400
right, at least these arguments 
need to be given. 

971
00:42:47,600 --> 00:42:52,160
And I will say someone needs to 
read Softbank's earnings like I 

972
00:42:52,160 --> 00:42:55,880
have though, because SoftBank 
does not remember when SoftBank 

973
00:42:55,880 --> 00:42:59,920
said they were going to be spend
$3 billion in open AI in back in

974
00:42:59,920 --> 00:43:01,680
February. 
SoftBank does not know how 

975
00:43:01,680 --> 00:43:03,880
that's working. 
Crystal Intelligence is yet to 

976
00:43:03,880 --> 00:43:07,920
pop the whole like there are 
straight up bullshit things 

977
00:43:07,920 --> 00:43:10,280
happening right and. 
Yeah, we're very interested in 

978
00:43:10,280 --> 00:43:13,200
SoftBank Rounder as well. 
We we agree with some of your 

979
00:43:13,200 --> 00:43:16,040
skepticism. 
But I I have a thing though with

980
00:43:16,040 --> 00:43:19,200
this, which is, putting aside 
how you think it may or not go, 

981
00:43:19,320 --> 00:43:21,920
what happen if Open AI doesn't 
raise their money? 

982
00:43:21,920 --> 00:43:24,400
They're dead. 
They are dead because they can't

983
00:43:24,400 --> 00:43:26,760
go public. 
And they also they had. 

984
00:43:26,760 --> 00:43:27,760
To you mean if they don't 
convert? 

985
00:43:27,760 --> 00:43:29,040
You mean if they don't convert 
to APB? 

986
00:43:29,040 --> 00:43:31,240
Yeah, if they don't convert, 
yeah, it's game over. 

987
00:43:31,400 --> 00:43:34,200
And I think open AI is a 
systemic risk because open AI 

988
00:43:34,200 --> 00:43:37,600
dies, Cor Weave dies. 
Crusoe has built nothing for 

989
00:43:37,600 --> 00:43:39,680
nothing. 
Oracle has sunk all this money 

990
00:43:39,680 --> 00:43:42,960
and JP Morgan Blue Owl primary 
data infrastructure has sunk all

991
00:43:42,960 --> 00:43:44,680
this money into data centres for
nobody. 

992
00:43:44,800 --> 00:43:47,360
They'll maybe resell them. 
I have some people at Oracle 

993
00:43:47,360 --> 00:43:50,760
telling me that they could find 
customers theoretically, but 

994
00:43:51,280 --> 00:43:54,200
this, all of this money is laid 
out for one company. 

995
00:43:54,520 --> 00:43:58,680
And if they don't convert and 
also SoftBank needs to get $22.5

996
00:43:58,680 --> 00:44:00,640
billion, they don't have that 
money. 

997
00:44:00,640 --> 00:44:03,400
Yes, they, they had a great 
quarter, whatever, but they 

998
00:44:03,400 --> 00:44:07,400
don't have the cash and they 
have a basic loan LTV thing 

999
00:44:07,560 --> 00:44:09,880
where it's like they can't go 
over 25%. 

1000
00:44:10,040 --> 00:44:12,560
They don't have even the space 
to do that. 

1001
00:44:12,640 --> 00:44:14,320
They're saying it will come out 
of vision fund. 

1002
00:44:14,440 --> 00:44:16,520
It's. 
People may lose them money, but 

1003
00:44:16,520 --> 00:44:19,080
there is some minimum price for 
like opening out. 

1004
00:44:19,080 --> 00:44:21,360
Like people see the value, 
they're producing revenue. 

1005
00:44:21,560 --> 00:44:23,880
So I just don't think, I just 
disagree that there's like some 

1006
00:44:23,880 --> 00:44:26,160
existential risk here. 
It's like, yes. 

1007
00:44:27,040 --> 00:44:28,760
There is. 
I just, they're, they're not 

1008
00:44:28,760 --> 00:44:31,440
going to die because like, even 
if it falls to 30 billion, 

1009
00:44:31,440 --> 00:44:33,960
because you only get this tiny 
sliver of it and the nonprofit 

1010
00:44:33,960 --> 00:44:37,280
owns tons of it, there's still 
more money that would come in at

1011
00:44:37,280 --> 00:44:41,040
some price to keep the thing. 
I mean, I think as long as it is

1012
00:44:41,040 --> 00:44:43,880
the leader in large language 
models and people are seeing 

1013
00:44:43,880 --> 00:44:45,920
tremendous value in language 
models, then yeah. 

1014
00:44:46,280 --> 00:44:48,720
Which I think the tremendous 
value. 

1015
00:44:49,080 --> 00:44:51,000
I think so that's that's the 
core area. 

1016
00:44:51,000 --> 00:44:54,000
We disagree that it is the 
number one app in the App Store 

1017
00:44:54,000 --> 00:44:57,720
is the main general purpose 
assistant to find it for me. 

1018
00:44:59,160 --> 00:45:00,920
But that was one. 
But that was one this. 

1019
00:45:01,280 --> 00:45:03,840
One time. 
It's a sustained #1 this isn't 

1020
00:45:03,840 --> 00:45:06,840
like a, you know, a podcaster 
getting the high up once and 

1021
00:45:06,840 --> 00:45:08,080
using the number over and over 
again. 

1022
00:45:08,080 --> 00:45:10,240
This is a sustained in the App 
Store. 

1023
00:45:10,480 --> 00:45:12,680
I mean worst case. 
I will I will say, Eric, like, 

1024
00:45:12,680 --> 00:45:14,520
like there are there are a 
warning as someone who covered 

1025
00:45:14,520 --> 00:45:17,240
Snapchat for many years, like 
you can, you can be #1 in the 

1026
00:45:17,240 --> 00:45:19,040
App Store and not really make 
that great of a business 

1027
00:45:19,040 --> 00:45:20,920
overall. 
So these things are somewhat if 

1028
00:45:20,920 --> 00:45:22,480
every. 
Customer loses you money, 

1029
00:45:22,720 --> 00:45:26,360
especially when you're burning 
money the moment anyone touches 

1030
00:45:26,360 --> 00:45:28,800
your software. 
Like that's the thing, gravity 

1031
00:45:28,800 --> 00:45:32,320
exists so sure you could. 
Well actually here's the thing. 

1032
00:45:32,520 --> 00:45:35,000
I get what your point is that if
they are the leader they are 

1033
00:45:35,120 --> 00:45:37,000
perceived as leader. 
They could keep raising money 

1034
00:45:37,280 --> 00:45:39,520
but why do people invest money 
in companies? 

1035
00:45:39,520 --> 00:45:43,200
It's so they can be an exit. 
The idea here is without the 

1036
00:45:43,200 --> 00:45:46,800
nonprofit conversion, open AI 
has no liquidity route other 

1037
00:45:46,800 --> 00:45:48,680
than what amounts to a Ponzi 
scheme. 

1038
00:45:48,880 --> 00:45:52,160
It just you are selling investor
money to other investors at 

1039
00:45:52,160 --> 00:45:54,080
increasing amounts I imagine 
maybe. 

1040
00:45:54,080 --> 00:45:56,400
An element and it's there are 
now all these secondary 

1041
00:45:56,400 --> 00:45:58,320
transactions, so it's like 
private companies. 

1042
00:45:58,320 --> 00:46:00,160
Or so that's the Ponzi scheme 
aspect of it. 

1043
00:46:00,600 --> 00:46:03,760
And I mean, how do you keep that
going if they could they be? 

1044
00:46:03,760 --> 00:46:06,600
Public, but you know, Stripe, 
all these these companies are 

1045
00:46:06,640 --> 00:46:09,960
selling there are there are 
exits involved like new people 

1046
00:46:09,960 --> 00:46:12,840
are buying old people. 
So you're just saying that Let's

1047
00:46:12,840 --> 00:46:17,240
just say the scheme requires 
money to keep entering the 

1048
00:46:17,240 --> 00:46:20,640
system, so that by exiting money
that's already entered the 

1049
00:46:20,640 --> 00:46:24,280
system, and if the money 
entering the system stops, then 

1050
00:46:24,280 --> 00:46:26,400
the business will the scheme in 
this case will die. 

1051
00:46:26,600 --> 00:46:29,920
That is, it's like if nobody 
buys, you know, Microsoft 

1052
00:46:29,920 --> 00:46:32,200
tomorrow, it'll fall to 0. 
Like this is just how stocks 

1053
00:46:32,200 --> 00:46:34,520
work, as long as they're. 
Microsoft makes more money than 

1054
00:46:34,520 --> 00:46:38,920
they lose billions of dollars. 
If there are transactions on the

1055
00:46:38,920 --> 00:46:41,800
secondary markets where big 
private companies are able to 

1056
00:46:41,800 --> 00:46:44,720
sell large chunks of stock to 
someone else, that's a price 

1057
00:46:44,880 --> 00:46:46,040
setting. 
Mechanism. 

1058
00:46:46,080 --> 00:46:50,160
Are you suggesting the open AI 
would continue selling stock if 

1059
00:46:50,160 --> 00:46:52,360
it couldn't go public? 
Is your suggestion just so I 

1060
00:46:52,360 --> 00:46:55,840
understand the open AI, they 
don't go, they don't convert, so

1061
00:46:55,840 --> 00:46:57,240
they are perpetually A 
nonprofit. 

1062
00:46:57,600 --> 00:46:59,640
You were suggesting that they 
would survive by just 

1063
00:46:59,760 --> 00:47:03,800
continually selling stock. 
No, They need some structure 

1064
00:47:03,800 --> 00:47:05,880
that's sustainable where they 
can say this is how it's going 

1065
00:47:05,880 --> 00:47:08,360
to work. 
This is where the profits go. 

1066
00:47:08,440 --> 00:47:10,920
This entity owns this percent of
profits. 

1067
00:47:10,920 --> 00:47:13,320
I'm investing in this. 
Like, I think there needs to be 

1068
00:47:13,320 --> 00:47:15,480
some resolution to that 
question, though. 

1069
00:47:15,480 --> 00:47:19,720
I think people clearly will have
a lot of energy to invest. 

1070
00:47:19,720 --> 00:47:22,840
Yeah, without that certainty. 
On the conversion aspect to it, 

1071
00:47:22,840 --> 00:47:25,880
Ed, the reason I, I have a hard 
time getting too worked up about

1072
00:47:25,880 --> 00:47:27,880
that is that we're not talking 
about some sort of technical 

1073
00:47:27,880 --> 00:47:30,040
breakthrough. 
We're not talking about AGI or, 

1074
00:47:30,040 --> 00:47:32,640
you know, them achieving some 
level of, you know, breakthrough

1075
00:47:32,640 --> 00:47:36,640
it in the language model. 
This is a political bureaucratic

1076
00:47:36,640 --> 00:47:39,000
transaction, right? 
This is about difficult it is to

1077
00:47:39,000 --> 00:47:40,640
convert. 
Sure, I've written lots of 

1078
00:47:40,640 --> 00:47:42,200
stories about this. 
I've written lots of stories 

1079
00:47:42,200 --> 00:47:44,120
about this. 
I mean, essentially their entire

1080
00:47:44,120 --> 00:47:47,160
fate rests in the hands of the 
AG in Delaware and California 

1081
00:47:47,560 --> 00:47:49,480
and to a smaller. 
Degree the IRS. 

1082
00:47:50,320 --> 00:47:52,480
Well, the IRS is, is downstream 
of all of this. 

1083
00:47:52,480 --> 00:47:55,000
I mean, this is mostly an AG 
decision and also Microsoft 

1084
00:47:55,000 --> 00:47:57,760
because it's about their larger 
contract over their, you know, 

1085
00:47:57,760 --> 00:47:59,360
relationship with open AI, 
right. 

1086
00:47:59,360 --> 00:48:02,160
I think Microsoft has a vested 
interest in ensuring that open 

1087
00:48:02,160 --> 00:48:04,520
AI doesn't collapse because 
their stock price is based on. 

1088
00:48:04,520 --> 00:48:05,720
Yeah, because their stock price 
is. 

1089
00:48:05,760 --> 00:48:07,800
Based on AP, they own all the 
infrastructure they. 

1090
00:48:07,800 --> 00:48:09,800
Just, but they also get, they 
also get cloud revenue from 

1091
00:48:09,800 --> 00:48:12,600
usage of Chachi BT I mean, if, 
if open AI disappear tomorrow, 

1092
00:48:12,600 --> 00:48:15,520
their story of an AI being an AI
stock would, would, would 

1093
00:48:15,520 --> 00:48:17,480
evaporate overnight. 
They can't sell copilot. 

1094
00:48:17,680 --> 00:48:19,520
You know, they, they, the 
Mustafa's team hasn't built 

1095
00:48:19,520 --> 00:48:21,240
shit. 
So they absolutely need open AI 

1096
00:48:21,240 --> 00:48:25,360
to exist. 
But what, what at least for the 

1097
00:48:25,360 --> 00:48:28,120
next couple of years. 
But look, I don't even know. 

1098
00:48:28,120 --> 00:48:30,600
I, I can't pretend I know how 
that one's going to going to 

1099
00:48:30,600 --> 00:48:33,080
play out on the, on the 
California AG front. 

1100
00:48:33,080 --> 00:48:35,320
I think there's going to be a 
huge fight with the nonprofits 

1101
00:48:35,880 --> 00:48:38,000
within the state and who even 
knows, you know, who becomes the

1102
00:48:38,000 --> 00:48:40,640
next governor of California? 
If that's Katie Porter, if it's 

1103
00:48:40,760 --> 00:48:43,360
I don't even know who else not 
not Kamala Harris. 

1104
00:48:43,560 --> 00:48:45,920
You know, I think one of the 
major things that will come up 

1105
00:48:45,920 --> 00:48:47,800
is how hard are you going to be 
on open AI? 

1106
00:48:48,040 --> 00:48:51,480
I think California's as a anti 
business state is something very

1107
00:48:51,480 --> 00:48:54,320
problematic to people here. 
And the idea that you would have

1108
00:48:54,320 --> 00:48:56,800
fucked over the most promising 
company by not allowing to 

1109
00:48:56,800 --> 00:49:00,520
convert into a for profit would 
be a huge problem. 

1110
00:49:00,520 --> 00:49:01,760
That. 
I mean actually we do that many.

1111
00:49:01,760 --> 00:49:05,000
Votes that you just like get big
enough and like the regulators 

1112
00:49:05,000 --> 00:49:06,760
like solve the problem problem 
for you and. 

1113
00:49:06,760 --> 00:49:09,120
I can't get that worked. 
Up even though I'm somewhat 

1114
00:49:09,120 --> 00:49:12,320
cynical about it. 
Same thing will happen for open 

1115
00:49:12,320 --> 00:49:14,360
AI. 
I would make I. 

1116
00:49:14,560 --> 00:49:16,840
Disagree with you, but I can see
how you got there. 

1117
00:49:17,040 --> 00:49:19,640
I wanna make a sort of different
point on the sort of 

1118
00:49:19,640 --> 00:49:22,920
overvaluation of AI startups 
that again parallels this like 

1119
00:49:22,920 --> 00:49:28,040
Uber trauma experience. 
It's just like SoftBank overpaid

1120
00:49:28,040 --> 00:49:30,840
for Uber, which effectively 
meant a bunch of Middle Eastern 

1121
00:49:30,840 --> 00:49:33,280
money and Saudi money overpaid 
for Uber. 

1122
00:49:33,480 --> 00:49:35,760
The valuation corrected and 
eventually went back up. 

1123
00:49:35,760 --> 00:49:39,280
Like a lot of the money that's 
flowing in now is these big like

1124
00:49:39,280 --> 00:49:42,880
Middle Eastern sovereign wealth.
And like some of my big question

1125
00:49:42,880 --> 00:49:45,560
to you like what is the more 
productive thing that this money

1126
00:49:45,560 --> 00:49:47,240
could go to? 
Like fundamentally what's 

1127
00:49:47,240 --> 00:49:50,000
happening is countries have made
a lot of money on oil. 

1128
00:49:50,240 --> 00:49:53,400
They don't have any better ideas
about what to do with the money.

1129
00:49:53,640 --> 00:49:55,880
So they throw it at AI and 
that's a. 

1130
00:49:55,880 --> 00:49:56,480
Hard. 
Line. 

1131
00:49:56,480 --> 00:49:59,120
Against the line. 
Like what I? 

1132
00:49:59,120 --> 00:50:01,320
Said There's a lot of very 
interesting cities that you 

1133
00:50:01,320 --> 00:50:04,400
could be building in Abu Dhabi 
and, and, and Riyadh right now. 

1134
00:50:04,400 --> 00:50:07,760
But I think you're overlooking. 
For a long quab is like mega 

1135
00:50:07,760 --> 00:50:10,240
city infrastructure projects, 
but they're they're basically 

1136
00:50:10,480 --> 00:50:13,160
saying, OK, the most productive 
thing going is an AI. 

1137
00:50:13,320 --> 00:50:16,560
And so worst case scenario, like
America is getting a bunch of 

1138
00:50:16,560 --> 00:50:21,040
mispriced money, oil money into 
our technology companies. 

1139
00:50:21,040 --> 00:50:23,360
And maybe it's slightly 
overpriced, but I, I just don't 

1140
00:50:23,360 --> 00:50:26,400
think the money is being burned.
Like it's largely American money

1141
00:50:26,400 --> 00:50:30,400
is being marked up by oil money.
And like, what's the tragedy 

1142
00:50:30,600 --> 00:50:33,760
from the American perspective? 
First of all, I think it's not 

1143
00:50:33,760 --> 00:50:36,000
an argument to say what else 
this should be spent on 

1144
00:50:36,000 --> 00:50:38,080
considering I already 
established that they're out of 

1145
00:50:38,080 --> 00:50:39,880
ideas. 
I do not know if there is 

1146
00:50:39,880 --> 00:50:42,680
another hyper growth market. 
I believe venture capital needs 

1147
00:50:42,680 --> 00:50:46,000
to actually look downstream and 
invest more money and seed and 

1148
00:50:46,000 --> 00:50:49,320
startups in general. 
Most of the money goes B and up 

1149
00:50:49,320 --> 00:50:51,840
if not C and up. 
It has historically done that. 

1150
00:50:51,920 --> 00:50:54,480
It's fucking stupid and 
antithetical to what Silicon 

1151
00:50:54,480 --> 00:50:57,160
Valley stands for. 
On top of that, I don't agree 

1152
00:50:57,160 --> 00:50:58,880
that most of this money is oil 
money. 

1153
00:50:58,880 --> 00:51:01,720
In fact, a remarkable amount of 
this money is coming through 

1154
00:51:01,720 --> 00:51:06,000
sources like Magnetar Capital or
NVIDIA or Microsoft or Amazon. 

1155
00:51:06,000 --> 00:51:09,200
Right, There's a lot of money. 
Flowing back Andreessen Horowitz

1156
00:51:09,200 --> 00:51:11,040
back to them, right? 
That's right. 

1157
00:51:11,120 --> 00:51:15,040
Okay, so you're just saying, So 
what is the thing you're saying 

1158
00:51:15,040 --> 00:51:17,680
that it's good this money's No, 
I'm saying like. 

1159
00:51:17,840 --> 00:51:20,720
This I'm saying valuation, even 
if these companies were all 

1160
00:51:20,720 --> 00:51:25,200
overvalued by 50%, I don't think
it's like a catastrophe like. 

1161
00:51:25,640 --> 00:51:28,960
Either I think it is a. 
With huge sums of money that 

1162
00:51:28,960 --> 00:51:30,800
don't know where else they want 
to invest it. 

1163
00:51:30,920 --> 00:51:34,680
It's companies that benefit from
the money coming back to them. 

1164
00:51:34,680 --> 00:51:36,360
Like you're saying like I agree 
with what you're saying. 

1165
00:51:36,360 --> 00:51:38,280
I just, I don't think that's 
Catas. 

1166
00:51:38,480 --> 00:51:41,560
There's a catastrophe. 
Evaluations fall so. 

1167
00:51:41,680 --> 00:51:43,120
I know, I know we're approaching
time. 

1168
00:51:43,120 --> 00:51:45,760
So I think I can actually answer
this fairly rigorously. 

1169
00:51:45,760 --> 00:51:49,240
So the catastrophe is the 
following in 2021 there was way 

1170
00:51:49,240 --> 00:51:52,440
too much money put in crap LP's 
got scared, start up money 

1171
00:51:52,440 --> 00:51:55,280
really slowed. 
It came back, it came back 

1172
00:51:55,280 --> 00:51:57,760
thanks to AI. 
The problem is that I actually I

1173
00:51:57,760 --> 00:52:00,720
think the money invested here is
stupider and the promises are 

1174
00:52:00,720 --> 00:52:02,840
more egregious, Right? 
Hear me out. 

1175
00:52:02,840 --> 00:52:03,640
No. 
No, no. 

1176
00:52:03,640 --> 00:52:05,480
Yeah, You're in this. 
This is a good ending, Yeah. 

1177
00:52:06,920 --> 00:52:08,760
Your feelings about what will 
won't happen. 

1178
00:52:08,960 --> 00:52:11,560
I don't think these companies 
are worth enough. 

1179
00:52:11,680 --> 00:52:13,720
I don't think that. 
I think they have the exit 

1180
00:52:13,720 --> 00:52:15,600
liquidity and think they're 
sustainable. 

1181
00:52:15,760 --> 00:52:18,840
As a result, a lot of venture 
capital is going to get burned, 

1182
00:52:18,840 --> 00:52:21,000
which means a lot of LP's are 
going to get burned, which is 

1183
00:52:21,000 --> 00:52:23,160
going to largely stymie the 
amount of money going into the 

1184
00:52:23,160 --> 00:52:27,000
startup ecosystem. 
On top of that, if NVIDIA has 

1185
00:52:27,040 --> 00:52:29,720
any problems with growth from 
here to oblivion, because they 

1186
00:52:29,720 --> 00:52:33,520
have to, it needs to be like 
what, 50 billion next quarter in

1187
00:52:33,520 --> 00:52:35,560
data center revenue? 
Even though data center revenue 

1188
00:52:35,560 --> 00:52:38,360
missed estimates last quarter, 
it needs to keep going what in 

1189
00:52:38,360 --> 00:52:41,600
three years we meant to be 
selling $120 billion of GPU's 

1190
00:52:41,600 --> 00:52:43,320
because that's what the market 
expects. 

1191
00:52:43,400 --> 00:52:46,600
All of these expectations have 
become load bearing. 

1192
00:52:46,840 --> 00:52:49,320
They become load bearing for 
Silicon Valley and they become 

1193
00:52:49,320 --> 00:52:52,920
load bearing for the markets. 
The problem is there is no way 

1194
00:52:52,920 --> 00:52:56,000
to match these expectations. 
So at some point, there is going

1195
00:52:56,000 --> 00:52:59,840
to be a nasty hangover. 
And my fear is it's going to hit

1196
00:52:59,840 --> 00:53:04,040
regular people in the markets, a
ton of retail money in NVIDIA 

1197
00:53:04,040 --> 00:53:05,800
too. 
And on top of that, it's going 

1198
00:53:05,800 --> 00:53:08,440
to stop start-ups getting money 
in general. 

1199
00:53:08,480 --> 00:53:10,880
And I think, you know, I think 
we need more money in angels 

1200
00:53:10,880 --> 00:53:12,440
anyway. 
I've always, always felt that 

1201
00:53:12,440 --> 00:53:14,320
way. 
It's like, I think that it's 

1202
00:53:14,360 --> 00:53:16,880
better to invest in earlier 
companies than otherwise. 

1203
00:53:17,280 --> 00:53:20,160
But I'm worried now that we're 
going to have a repeat of 

1204
00:53:20,160 --> 00:53:24,240
2022-2023 ass layoffs. 
Fuck all money for startups. 

1205
00:53:24,400 --> 00:53:27,720
A lot of cynicism, and I don't 
mean cynicism like me, I mean 

1206
00:53:27,720 --> 00:53:30,600
just straight up, I don't want 
to put money in tech and when 

1207
00:53:30,600 --> 00:53:33,720
tech doesn't have a hyper growth
thing after AI, I'm terrified 

1208
00:53:33,720 --> 00:53:36,360
that it's going to just start 
slashing tech valuations across 

1209
00:53:36,360 --> 00:53:37,960
the board. 
However you feel about 

1210
00:53:37,960 --> 00:53:41,320
generative AI, my arguments are 
founded on like real things. 

1211
00:53:41,320 --> 00:53:44,200
And I mentioned cognition, for 
example, the evaluation. 

1212
00:53:44,200 --> 00:53:46,520
It's like, how's that? 
And how does any of this? 

1213
00:53:46,520 --> 00:53:48,360
And does open AI really go 
public? 

1214
00:53:48,360 --> 00:53:51,280
Does anthropic go public? 
Do you think that their S 1 is 

1215
00:53:51,280 --> 00:53:53,240
going to look like anything 
other than a Jackson Pollock 

1216
00:53:53,240 --> 00:53:56,080
painting? 
Like it's, it's rough out there 

1217
00:53:56,200 --> 00:54:00,440
and I'm actually, I don't know 
how people are so bearish right 

1218
00:54:00,440 --> 00:54:02,560
now. 
I don't know how I get why 

1219
00:54:02,560 --> 00:54:06,440
people aren't like completely 
insane and swagged up like me, 

1220
00:54:06,440 --> 00:54:11,320
but I don't, I don't get why 
more people aren't worried about

1221
00:54:11,320 --> 00:54:14,440
it because this doesn't even 
match previous things. 

1222
00:54:14,440 --> 00:54:17,320
And I'm just, I have real 
concerns and the people who are 

1223
00:54:17,320 --> 00:54:20,760
going to get hurt are founders 
and real people and who have 

1224
00:54:20,760 --> 00:54:23,160
invested in the market. 
It's got their downstream 

1225
00:54:23,160 --> 00:54:25,840
effects are rough. 
The Silicon Valley consensus is 

1226
00:54:25,840 --> 00:54:28,400
that Angel and seed rounds are 
overvalued. 

1227
00:54:28,400 --> 00:54:30,560
Now, maybe you disagree, but 
it's just funny to me to hear it

1228
00:54:30,560 --> 00:54:31,960
from you. 
I think it's. 

1229
00:54:32,080 --> 00:54:35,080
I think it's. 
Far more, I think it's far more 

1230
00:54:35,080 --> 00:54:39,120
reasonable to overvalue at the 
seed other than fucking thinking

1231
00:54:39,120 --> 00:54:41,120
machines. 
And that's a super safe 

1232
00:54:41,120 --> 00:54:42,800
intelligence. 
I think those are ludicrous. 

1233
00:54:43,080 --> 00:54:46,560
But I think it's OK to take bets
at the seed and the Series A 

1234
00:54:46,560 --> 00:54:48,480
stage. 
That's that's fine. 

1235
00:54:48,480 --> 00:54:50,240
Like, come on. 
Like, yeah, you take the piss a 

1236
00:54:50,240 --> 00:54:53,320
bit there because you have hopes
and you get to these overblown 

1237
00:54:53,320 --> 00:54:56,520
valuations. 
Open AI stage, anthropic stage. 

1238
00:54:56,520 --> 00:54:58,920
They should be sustainable and I
think that's a reasonable thing 

1239
00:54:58,920 --> 00:54:59,920
to ask for. 
All right. 

1240
00:54:59,920 --> 00:55:01,320
Well, look, I mean, the 
beautiful thing about where 

1241
00:55:01,320 --> 00:55:03,600
we're at right now is that we 
don't know who's right. 

1242
00:55:04,160 --> 00:55:05,400
We have no idea what's going to 
happen. 

1243
00:55:06,600 --> 00:55:08,040
The future remains. 
To be written. 

1244
00:55:08,600 --> 00:55:11,960
Yeah, if it is written, it could
be written by AI and we, we know

1245
00:55:11,960 --> 00:55:14,440
that it's very sycophant, think 
it will make us feel very smart 

1246
00:55:14,440 --> 00:55:16,280
about what we asked it. 
But Ed, thank you so much for 

1247
00:55:16,280 --> 00:55:18,120
joining. 
I have a feeling we should do 

1248
00:55:18,120 --> 00:55:21,280
this more often. 
I would love to thanks for I 

1249
00:55:21,280 --> 00:55:23,560
know this has been quite I've 
really had a good time. 

1250
00:55:23,800 --> 00:55:25,320
No, we. 
Coming a little. 

1251
00:55:26,080 --> 00:55:28,640
Bit I actually pay for new coat 
like I have quite sometimes. 

1252
00:55:28,800 --> 00:55:31,320
We enjoy your sub stack. 
So yeah, thank you very much. 

1253
00:55:31,440 --> 00:55:32,040
Thank. 
You so much. 

1254
00:55:32,440 --> 00:55:33,000
Yeah. 
All right. 

1255
00:55:33,000 --> 00:55:33,200
Thanks.